1. A barrier-free VR teaching resource color optimization method for people with abnormal color vision is characterized by comprising the following steps:

(1) constructing a color cognition factor system, namely constructing a color factor system of barrier-free VR teaching resources from the aspects of color rules, visual physiological structures and visual psychology feelings; respectively expressing the characteristics of the three types of color factors by adopting different parameters; acquiring color samples of a VR scene, constructing a color sample library, establishing an expert knowledge base of VR teaching resource color quality, designing various types of tasks, and analyzing color cognitive difference among different crowds;

(2) constructing a color evaluation model, screening alternative factors of the color evaluation model, describing the characteristics of an input color scheme, and determining the weight of the factors; constructing color sample libraries of different disciplines, styles and types, generating color labels of the color samples, and extracting a color scheme of the teaching resources; designing a color evaluation regressor, and comprehensively evaluating the color quality prediction of the learner by adopting various indexes by using an Adaboost integrated learning model;

(3) generating a color scheme optimization model, supporting color evaluation generation facing different crowds, constructing a general optimal color selection scheme, and completing a customized color selection model facing specific color vision abnormal crowds; dividing the color pattern spots into different areas by using a Voronoi method, establishing a color selection model facing the cognition of people with abnormal color vision, and designing a barrier-free VR teaching resource color scheme; screening out the optimal color scheme to generate a candidate generation scheme for optimizing the color selection; and establishing an optimized color selection model to realize barrier-free optimization of the original VR teaching resource color scheme.

2. The method for optimizing colors of barrier-free VR instructional resources for people with color vision disorders as claimed in claim 1, wherein the step (1) of constructing a color cognition factor system specifically comprises:

(1-1) constructing a color factor system, and extracting color rule series factors from the aspects of color matching, color discrimination and color significance; constructing a physiological structure factor according to the aspects of the color blindness type, the color weakness type and the severity, forming a color psychosensory factor according to the aspects of color harmony, color cooling and warming and color enthusiasm, and constructing a color factor system of the barrier-free VR teaching resource;

(1-1-1) extracting color rule series factors, summarizing color design knowledge of VR teaching resources from aspects of color self rules, color expression types, color semantics, object forms and distribution, and extracting color matching, distinguishing degree, uniformity, significance and internal semantic relation series factors;

(1-1-2) constructing physiological structure series factors of people with abnormal color vision, analyzing the difference of a tested object in visual physiological perception, and constructing color vision physiological structure series factors of VR teaching resources aiming at the difference of color vision abnormal types, color blindness types, color weakness types and color vision abnormal severity;

(1-1-3) extracting color psychosensory series factors, extracting color harmony, color cooling and heating, color active/negative emotion series factors aiming at the integral visual perception of the VR teaching resource color, and judging the weight of the color harmony and the color emotional psychosensory factors according to the color quality degree influencing the barrier-free VR teaching resource;

(1-2) carrying out color factor characterization expression, and obtaining the characteristics of perception difference, uniformity and variation trend of different colors according to a CIEDE2000 color difference model; expressing visual physiological structure factors by using a color spectrum perception function or a color weakness spectrum perception curve; calculating the difference values of the hue, the saturation and the brightness to obtain color harmony characteristics, and calculating color emotional characteristics by using the values of color cold and warm, lightness, positivity/negativity and the hue angle;

(1-2-1) carrying out characteristic expression on color rule series factors, and acquiring perception difference characteristics among different colors by using three components of lightness, chroma and hue according to a CIEDE2000 color difference model; calculating the distance between adjacent colors to obtain the uniformity characteristic of perception difference; extracting the variation trend characteristics of the color by using the monotonicity of the color;

(1-2-2) performing characteristic expression on the physiological series factors of the color vision anomaly crowd, and calculating a projection plane of the color vision anomaly crowd in a degraded color gamut by using a spectrum perception function of colors in the color blindness crowd; calculating the interpolation position of the color on the color weakness spectrum perception curve, and expressing the physiological structure factors of color blindness and color weakness vision;

(1-2-3) performing characteristic expression on the color psychology feeling series factors, calculating the difference values of hue, saturation and brightness in a CIELab color space, and acquiring color harmony characteristics reflecting color visual psychology feelings; calculating color emotional characteristics by using the values of color cold and warm, light and heavy, positive/negative and the hue angle in the color space;

(1-3) color quality cognition, namely selecting different color dimension indexes and establishing an expert knowledge base of VR teaching resource color quality; designing various types of cognitive tasks around the theme matching of teaching resources; different cognition tasks are completed by the group of normal and abnormal contrast groups of tissue color vision, and the color cognition difference among different groups of people is analyzed;

(1-3-1) generating an expert knowledge base of color quality, and establishing an expert knowledge base of VR teaching resource color quality according to color discrimination, color uniformity, color semantics, color selection paths, color monotonicity, color geometrical morphology, color space distribution, color harmony and color emotion dimensionality;

(1-3-2) designing a color cognitive task, and designing and positioning, identifying, comparing, sequencing, associating and recalling various cognitive tasks around hue difference, hue quantity, color distance, color area, color distribution, color internal semantic relation and color and teaching resource theme matching factors;

(1-3-3) analyzing color cognition difference, organizing the crowd with normal and abnormal color vision to complete corresponding color cognition tasks, and analyzing the color cognition difference among different crowds by collecting the learning behavior indexes of the tested object, such as accuracy, completion efficiency, watching duration, watching frequency and eye jump times eye movement indexes.

3. The method for optimizing colors of barrier-free VR instructional resources for people with color vision disorders as claimed in claim 1, wherein said step (2) of building a color evaluation model specifically comprises:

(2-1) screening factors which have obvious influence on the cognition of corresponding crowds by the characteristic combination of the color evaluation model to serve as alternative factors of the color evaluation model; selecting alternative factors with high association degree by using a machine learning method, and describing the characteristics of the input color scheme; determining the weight of the color evaluation model factor by applying a machine learning method;

(2-1-1) selecting alternative factors, and screening factors which have obvious influence on cognition of corresponding types of crowds from three series factors of color rules, color vision physiological structures and color vision psychological feelings according to the types of color schemes and cognitive differences facing the crowds to serve as the alternative factors of a color evaluation model;

(2-1-2) performing model feature description, selecting alternative factors with high association degree by using a machine learning method from the brightness, hue, lightness, saturation and harmonicity dimensions of VR teaching resource colors according to the cognitive mechanism difference of different types of people on the colors, and describing the features of an input color scheme;

(2-1-3) determining factor weight, determining influence weight of characteristic factors of a color evaluation model by applying a machine learning method according to the category to which the VR teaching resource color scheme belongs, and highlighting monotonicity, discrimination, uniformity, significance and difference of harmonic factors of the color scheme;

(2-2) extracting a color scheme, collecting color samples of the existing VR scene, and constructing color sample libraries of different disciplines, styles and types; generating a color label of the sample based on the prediction result of the regressor and the actual category of the sample; extracting color features of the picture by using the color histogram to obtain a complete color scheme;

(2-2-1) constructing a color sample library, collecting different scenes from the existing VR games and VR teaching resources, collecting color samples by using a spectrophotometry color meter, color-taking software, network crawler hardware equipment or software, and constructing color sample libraries of different subjects, different styles and different types;

(2-2-2) generating color sample labels, selecting a GE-SMOTE method to preprocess a data set aiming at the condition of color distribution imbalance in VR teaching resources, then constructing a mixing matrix by combining a DataBoost-IM method, and generating the color labels of the samples based on the prediction result of the regressor and the actual types of the samples;

(2-2-3) extracting a color scheme, namely extracting picture color features of VR teaching resources by using a color histogram, setting a CIELab threshold value of each color, and tracking the boundary of the color in an image by adopting binarization processing to obtain a corresponding image spot area; obtaining a complete color scheme in a VR teaching resource picture through multiple iterations;

(2-3) predicting color quality, designing a color evaluation regressor, and creating a learner combination consisting of a plurality of base learners by using a series connection and parallel connection mode; an Adaboost ensemble learning model is used, and the accuracy of the color sample training result is improved by combining a framework of a cascade learner; comprehensively evaluating the color quality predicted by the learner by adopting various indexes;

(2-3-1) designing a color evaluation regressor, based on a CIELab color model, extracting color patches of VR teaching resources by using a k-means method, adopting a KNN, logistic regression and decision tree classical regressor model as a base learner, and creating a learner combination consisting of a plurality of base learners by using a series connection and parallel connection mode;

(2-3-2) designing an ensemble learning model, adopting an Adaboost ensemble learning model, combining a framework of a cascade learner, comprehensively considering a historical judgment result and a current judgment result of a color scheme, and adding an auxiliary judgment function to improve the accuracy of a color sample training result;

(2-3-3) color quality prediction, namely comprehensively evaluating the color quality predicted by the learners by adopting average absolute error, root mean square error and goodness-of-fit indexes, comparing the prediction performances of a plurality of learners, selecting a strong learner with the optimal prediction performance, and realizing the color quality prediction of VR teaching resources for different crowds.

4. The method for optimizing colors of barrier-free VR instructional resources for people with color vision disorders as claimed in claim 1, wherein the step (3) of generating color scheme optimization model specifically comprises:

(3-1) organizing different people to score according to a Likter scale by using a color optimal selection model of VR teaching resource colors, and supporting color evaluation generation facing different crowds; taking the color evaluation of the crowd with abnormal color vision as a constraint condition, and taking the color evaluation of the normal crowd as a target function to construct a general optimal color selection scheme; considering the cognitive characteristics of specific color vision abnormal crowds, constructing a customized color selection model;

(3-1-1) color evaluation generation, wherein colors of VR teaching resources are associated with scenes with different geometric forms and different spatial layouts to form a series of VR teaching resources, color vision anomaly and normal personnel are organized to be scored according to a Likter scale, and color evaluation generation facing different crowds is supported;

(3-1-2) constructing a color selection scheme of a general VR teaching resource by taking the color evaluation of the color vision abnormal crowd as a constraint condition and the color evaluation of the normal crowd as a target function, and solving the optimal color scheme relative to the normal crowd in the color series scheme meeting the cognition of the color vision abnormal crowd;

(3-1-3) customizing a color selection model, taking the cognitive characteristics of specific color vision abnormal crowds into consideration, taking the color evaluation of barrier-free VR teaching resources of color vision abnormal personnel as a target function, constructing the customized color selection model, and solving a color scheme which maximally meets the cognition of the specific color vision abnormal crowds from a color space;

(3-2) constructing a barrier-free VR teaching resource color scheme, and aiming at the color scheme of the existing sample, dividing color pattern spots into different areas by using a Voronoi method; establishing a color selection scheme facing the cognition of people with abnormal color vision by using a color rule and an evaluation result as a constraint condition and a target function; introducing a firefly algorithm, and optimizing the construction of a barrier-free VR teaching resource color scheme;

(3-2-1) image segmentation, namely analyzing the distribution of different colors in an image on a CIELab space by utilizing nuclear density according to a color sample library or a color scheme of an existing VR teaching resource, merging similar color areas by adopting a spatial clustering method, calculating the gravity center of a region polygon, segmenting the image by utilizing a Voronoi diagram, and dividing a color pattern into different areas;

(3-2-2) establishing a barrier-free color selection scheme, combining the Voronoi subdivision-based patches into different color schemes according to a barrier-free color design rule, and establishing a barrier-free VR teaching resource color selection scheme facing the cognition of people with abnormal color vision by using the color rules and the evaluation results as constraint conditions and objective functions;

(3-2-3) constructing an optimized color scheme, introducing a firefly algorithm, generating an optimal solution by using a neural network algorithm in the color selection scheme, introducing the position of a candidate region in the optimal descrambling color scheme, improving a local search position updating formula, overcoming the defect of easy generation of a local optimal solution, and optimizing the construction of a barrier-free VR teaching resource color scheme;

(3-3) carrying out barrier-free transformation on the original VR teaching resource color scheme, and screening out the optimal color scheme according to the logic structure of the color; generating a candidate generation scheme of color selection optimization by using Voronoi subdivision; establishing an optimized color selection model taking the similarity as a constraint condition and the color evaluation difference as a target function, and realizing barrier-free transformation of the original VR teaching resource color scheme;

(3-3-1) selecting a color scheme, and determining the category of the color scheme according to the logic structure of the original VR teaching resource color, wherein the category comprises qualitative color, sequential color and colors at two ends; selecting a series of schemes which accord with corresponding rules in a color system or a color space by using the design rules of the color schemes, and further screening out the optimal color scheme;

(3-3-2) generating a color candidate optimization scheme, wherein Voronoi is adopted to subdivide different color patches by taking the spatial distribution of the original VR teaching resource color scheme as a reference based on a CIELab color space, each generated Voronoi subspace is a candidate area for optimizing and selecting colors of the color scheme, and the combination of different areas can form the candidate scheme for color optimization;

(3-3-3) selecting an optimization scheme, calculating the similarity of the VR teaching resource color scheme based on the color series factors, and judging the difference of the color style before and after optimization through comparing the similarity; and establishing an optimized color selection model taking the similarity as a constraint condition and the color evaluation difference as a target function, and realizing barrier-free optimization of the original VR teaching resource color.

Background

Virtual teaching resources generated based on Virtual Reality (VR) technology have a highly-impulsive visual effect, and can deepen the memory and understanding of learners on knowledge contents. However, people with abnormal color vision cannot perceive the color model of the teaching resources like normal people, and the color scheme of the virtual teaching resources is difficult to effectively perceive. In view of the design and development of virtual teaching resources, the demands of people with normal color vision are met as targets, and the virtual teaching resources are not friendly to people with abnormal color vision in the aspects of color matching schemes and visual transmission, so that the people are confused, and unfair results are caused in the actual teaching process. The design and development process of the existing VR teaching resource color scheme is optimized, the barrier-free application of VR teaching resources is realized, and a wide application prospect is brought to the education field.

At present, the VR teaching resource has the following problems in the aspect of color scheme construction: (1) as an emerging application field, the demands of people with abnormal color vision are less considered in resource design and use by the industry and researchers; (2) with the wide application of VR teaching resources, the demand for barrier-free VR teaching resources is increased rapidly, and a universal color evaluation model is lacked; (3) how to reform transform the current color scheme, generate accessible VR teaching resource will save a large amount of manpowers, cost.

Disclosure of Invention

Aiming at the defects or the improvement requirements of the prior art, the invention provides a barrier-free VR teaching resource color optimization method for people with abnormal color vision, and provides a new method and path for the construction, evaluation, optimization and modification of a barrier-free VR teaching resource color scheme.

The object of the invention is achieved by the following technical measures.

A barrier-free VR teaching resource color optimization method for people with abnormal color vision comprises the following steps:

(1) constructing a color cognition factor system, namely constructing a color factor system of barrier-free VR teaching resources from the aspects of color rules, visual physiological structures and visual psychology feelings; respectively expressing the characteristics of the three types of color factors by adopting different parameters; acquiring color samples of a VR scene, constructing a color sample library, establishing an expert knowledge base of VR teaching resource color quality, designing various types of tasks, and analyzing color cognitive difference among different crowds;

(2) constructing a color evaluation model, screening alternative factors of the color evaluation model, describing the characteristics of an input color scheme, and determining the weight of the factors; constructing color sample libraries of different disciplines, styles and types, generating color labels of the color samples, and extracting a color scheme of the teaching resources; designing a color evaluation regressor, and comprehensively evaluating the color quality prediction of the learner by adopting various indexes by using an Adaboost integrated learning model;

(3) generating a color scheme optimization model, supporting color evaluation generation facing different crowds, constructing a general optimal color selection scheme, and completing a customized color selection model facing specific color vision abnormal crowds; dividing the color pattern spots into different areas by using a Voronoi method, establishing a color selection model facing the cognition of people with abnormal color vision, and designing a barrier-free VR teaching resource color scheme; screening out the optimal color scheme to generate a candidate generation scheme for optimizing the color selection; and establishing an optimized color selection model to realize barrier-free optimization of the original VR teaching resource color scheme.

In the above technical solution, "the color cognition factor system construction" in the step (1) specifically includes:

(1-1) constructing a color factor system, and extracting color rule series factors from the aspects of color matching, color discrimination and color significance; constructing a physiological structure factor according to the aspects of the color blindness type, the color weakness type and the severity, forming a color psychosensory factor according to the aspects of color harmony, color cooling and warming and color enthusiasm, and constructing a color factor system of the barrier-free VR teaching resource;

(1-1-1) extracting color rule series factors, summarizing color design knowledge of VR teaching resources from aspects of color self rules, color expression types, color semantics, object forms and distribution, and extracting color matching, distinguishing degree, uniformity, significance and internal semantic relation series factors;

(1-1-2) constructing a physiological structure series factor of a crowd with abnormal color vision, analyzing the difference of a tested object in visual physiological perception, and constructing a VR teaching resource color physiological structure series factor aiming at the color vision abnormal type (color blindness/color weakness), the color blindness type (red blindness, green blindness and the like), the color weakness type (red weakness, green weakness and the like) and the color vision abnormal severity difference;

(1-1-3) extracting color psychosensory series factors, extracting color harmony, color cooling and heating, color active/negative emotion series factors aiming at the integral visual perception of the VR teaching resource color, and judging the weight of the color harmony and the color emotional psychosensory factors according to the color quality degree influencing the barrier-free VR teaching resource;

(1-2) carrying out color factor characterization expression, and obtaining the characteristics of perception difference, uniformity and variation trend of different colors according to a CIEDE2000 color difference model; expressing visual physiological structure factors by using a color spectrum perception function or a color weakness spectrum perception curve; calculating the difference values of the hue, the saturation and the brightness to obtain color harmony characteristics, and calculating color emotional characteristics by using the values of color cold and warm, lightness, positivity/negativity and the hue angle;

(1-2-1) carrying out characteristic expression on color rule series factors, and acquiring perception difference characteristics among different colors by using three components of lightness, chroma and hue according to a CIEDE2000 color difference model; calculating the distance between adjacent colors to obtain the uniformity characteristic of perception difference; extracting the variation trend characteristics of the color by using the monotonicity of the color;

(1-2-2) performing characteristic expression on the physiological series factors of the color vision anomaly crowd, and calculating a projection plane of the color vision anomaly crowd in a degraded color gamut by using a spectrum perception function of colors in the color blindness crowd; calculating the interpolation position of the color on the color weakness spectrum perception curve, and expressing the physiological structure factors of color blindness and color weakness vision;

(1-2-3) performing characteristic expression on the color psychology feeling series factors, calculating the difference values of hue, saturation and brightness in a CIELab color space, and acquiring color harmony characteristics reflecting color visual psychology feelings; calculating color emotional characteristics by using the values of color cold and warm, light and heavy, positive/negative and the hue angle in the color space;

(1-3) color quality cognition, namely selecting different color dimension indexes and establishing an expert knowledge base of VR teaching resource color quality; designing various types of cognitive tasks around the theme matching of teaching resources; different cognition tasks are completed by the group of normal and abnormal contrast groups of tissue color vision, and the color cognition difference among different groups of people is analyzed;

(1-3-1) generating an expert knowledge base of color quality, and establishing an expert knowledge base of VR teaching resource color quality according to color discrimination, color uniformity, color semantics, color selection paths, color monotonicity, color geometrical morphology, color space distribution, color harmony and color emotion dimensionality;

(1-3-2) designing a color cognitive task, and designing and positioning, identifying, comparing, sequencing, associating and recalling various cognitive tasks around hue difference, hue quantity, color distance, color area, color distribution, color internal semantic relation and color and teaching resource theme matching factors;

(1-3-3) analyzing color cognition difference, organizing the crowd with normal and abnormal color vision to complete corresponding color cognition tasks, and analyzing the color cognition difference among different crowds by collecting the learning behavior indexes of the tested object, such as accuracy, completion efficiency, watching duration, watching frequency and eye jump times eye movement indexes.

In the above technical solution, "constructing a color evaluation model" in step (2) specifically includes:

(2-1) screening factors which have obvious influence on the cognition of corresponding crowds by the characteristic combination of the color evaluation model to serve as alternative factors of the color evaluation model; selecting alternative factors with high association degree by using a machine learning method, and describing the characteristics of the input color scheme; determining the weight of the color evaluation model factor by applying a machine learning method;

(2-1-1) selecting alternative factors, and screening factors which have obvious influence on cognition of corresponding types of crowds from three series factors of color rules, color vision physiological structures and color vision psychological feelings according to the types of color schemes and cognitive differences facing the crowds to serve as the alternative factors of a color evaluation model;

(2-1-2) performing model feature description, selecting alternative factors with high association degree by using a machine learning method from the brightness, hue, lightness, saturation and harmonicity dimensions of VR teaching resource colors according to the cognitive mechanism difference of different types of people on the colors, and describing the features of an input color scheme;

(2-1-3) determining factor weight, determining influence weight of characteristic factors of a color evaluation model by applying a machine learning method according to the category (such as qualitative color, sequential color or two-end color) to which the VR teaching resource color scheme belongs, and highlighting monotonicity, discrimination, uniformity, significance and difference of harmonic factors of the color scheme;

(2-2) extracting a color scheme, collecting color samples of the existing VR scene, and constructing color sample libraries of different disciplines, styles and types; generating a color label of the sample based on the prediction result of the regressor and the actual category of the sample; extracting color features of the picture by using the color histogram to obtain a complete color scheme;

(2-2-1) constructing a color sample library, collecting different scenes from the existing VR games and VR teaching resources, collecting color samples by using a spectrophotometry color meter, color-taking software, network crawler hardware equipment or software, and constructing color sample libraries of different subjects, different styles and different types;

(2-2-2) generating color sample labels, selecting a GE-SMOTE method to preprocess a data set aiming at the condition of color distribution imbalance in VR teaching resources, then constructing a mixing matrix by combining a DataBoost-IM method, and generating the color labels of the samples based on the prediction result of the regressor and the actual types of the samples;

(2-2-3) extracting a color scheme, namely extracting picture color features of VR teaching resources by using a color histogram, setting a CIELab threshold value of each color, and tracking the boundary of the color in an image by adopting binarization processing to obtain a corresponding image spot area; obtaining a complete color scheme in a VR teaching resource picture through multiple iterations;

(2-3) predicting color quality, designing a color evaluation regressor, and creating a learner combination consisting of a plurality of base learners by using a series connection and parallel connection mode; an Adaboost ensemble learning model is used, and the accuracy of the color sample training result is improved by combining a framework of a cascade learner; comprehensively evaluating the color quality predicted by the learner by adopting various indexes;

(2-3-1) designing a color evaluation regressor, based on a CIELab color model, extracting color patches of VR teaching resources by using a k-means method, adopting a KNN, logistic regression and decision tree classical regressor model as a base learner, and creating a learner combination consisting of a plurality of base learners by using a series connection and parallel connection mode;

(2-3-2) designing an ensemble learning model, adopting an Adaboost ensemble learning model, combining a framework of a cascade learner, comprehensively considering a historical judgment result and a current judgment result of a color scheme, and adding an auxiliary judgment function to improve the accuracy of a color sample training result;

(2-3-3) color quality prediction, namely comprehensively evaluating the color quality predicted by the learners by adopting average absolute error, root mean square error and goodness-of-fit indexes, comparing the prediction performances of a plurality of learners, selecting a strong learner with the optimal prediction performance, and realizing the color quality prediction of VR teaching resources for different crowds.

In the above technical solution, the "color scheme optimization model generation" in step (3) specifically includes:

(3-1) organizing different people to score according to a Likter scale by using a color optimal selection model of VR teaching resource colors, and supporting color evaluation generation facing different crowds; taking the color evaluation of the crowd with abnormal color vision as a constraint condition, and taking the color evaluation of the normal crowd as a target function to construct a general optimal color selection scheme; considering the cognitive characteristics of specific color vision abnormal crowds, constructing a customized color selection model;

(3-1-1) color evaluation generation, wherein colors of VR teaching resources are associated with scenes with different geometric forms and different spatial layouts to form a series of VR teaching resources, color vision anomaly and normal personnel are organized to be scored according to a Likter scale, and color evaluation generation facing different crowds is supported;

(3-1-2) constructing a color selection scheme of a general VR teaching resource by taking the color evaluation of the color vision abnormal crowd as a constraint condition and the color evaluation of the normal crowd as a target function, and solving the optimal color scheme relative to the normal crowd in the color series scheme meeting the cognition of the color vision abnormal crowd;

(3-1-3) customizing a color selection model, taking the cognitive characteristics of specific color vision abnormal crowds into consideration, taking the color evaluation of barrier-free VR teaching resources of color vision abnormal personnel as a target function, constructing the customized color selection model, and solving a color scheme which maximally meets the cognition of the specific color vision abnormal crowds from a color space;

(3-2) constructing a barrier-free VR teaching resource color scheme, and aiming at the color scheme of the existing sample, dividing color pattern spots into different areas by using a Voronoi method; establishing a color selection scheme facing the cognition of people with abnormal color vision by using a color rule and an evaluation result as a constraint condition and a target function; introducing a firefly algorithm, and optimizing the construction of a barrier-free VR teaching resource color scheme;

(3-2-1) image segmentation, namely analyzing the distribution of different colors in an image on a CIELab space by utilizing nuclear density according to a color sample library or a color scheme of an existing VR teaching resource, merging similar color areas by adopting a spatial clustering method, calculating the gravity center of a region polygon, segmenting the image by utilizing a Voronoi diagram, and dividing a color pattern into different areas;

(3-2-2) establishing a barrier-free color selection scheme, combining the Voronoi subdivision-based patches into different color schemes according to a barrier-free color design rule, and establishing a barrier-free VR teaching resource color selection scheme facing the cognition of people with abnormal color vision by using the color rules and the evaluation results as constraint conditions and objective functions;

(3-2-3) constructing an optimized color scheme, introducing a firefly algorithm, generating an optimal solution by using a neural network algorithm in the color selection scheme, introducing the position of a candidate region in the optimal descrambling color scheme, improving a local search position updating formula, overcoming the defect of easy generation of a local optimal solution, and optimizing the construction of a barrier-free VR teaching resource color scheme;

(3-3) carrying out barrier-free transformation on the original VR teaching resource color scheme, and screening out the optimal color scheme according to the logic structure of the color; generating a candidate generation scheme of color selection optimization by using Voronoi subdivision; establishing an optimized color selection model taking the similarity as a constraint condition and the color evaluation difference as a target function, and realizing barrier-free transformation of the original VR teaching resource color scheme;

(3-3-1) selecting a color scheme, and determining the type of the color scheme, such as qualitative color, sequential color, colors at two ends and the like, according to the logic structure of the original VR teaching resource color; selecting a series of schemes which accord with corresponding rules in a color system or a color space by using the design rules of the color schemes, and further screening out the optimal color scheme;

(3-3-2) generating a color candidate optimization scheme, wherein Voronoi is adopted to subdivide different color patches by taking the spatial distribution of the original VR teaching resource color scheme as a reference based on a CIELab color space, each generated Voronoi subspace is a candidate area for optimizing and selecting colors of the color scheme, and the combination of different areas can form the candidate scheme for color optimization;

(3-3-3) selecting an optimization scheme, calculating the similarity of the VR teaching resource color scheme based on the color series factors, and judging the difference of the color style before and after optimization through comparing the similarity; and establishing an optimized color selection model taking the similarity as a constraint condition and the color evaluation difference as a target function, and realizing barrier-free optimization of the original VR teaching resource color.

Compared with the prior art, the barrier-free VR teaching resource color optimization method for the crowd with abnormal color vision has the beneficial effects that:

constructing a color factor system of barrier-free VR teaching resources from the aspects of color rules, visual physiological structures and visual psychosocial feelings, and respectively expressing the characteristics of the factors; establishing an expert knowledge base of color quality, designing various tasks, and analyzing the color cognitive difference among different crowds. Screening alternative factors of the color evaluation model, selecting characteristic factors describing a color scheme, and determining the weight of the factors; constructing a color sample library, generating a color label, and extracting a color scheme of the teaching resource; and comprehensively evaluating the color quality prediction of the learner by using an Adaboost ensemble learning model. Supporting color evaluation generation, constructing a general optimal color selection scheme, and customizing a color selection model; segmenting into color patches by using a Voronoi method, and designing a barrier-free VR teaching resource color scheme; the barrier-free transformation of the original VR teaching resource color scheme is realized. With the gradual popularization of the 5G network environment, the application of VR in teaching is more and more popular, and the demand of providing barrier-free VR teaching resource color schemes for students/teachers with abnormal color vision is more and more urgent. The invention is beneficial to non-professional persons to design and optimize VR teaching resources, achieves the universality of the color scheme and meets the fairness requirement of education resources.

Drawings

Fig. 1 is a flowchart of a barrier-free VR teaching resource color optimization method for people with color vision abnormalities in the embodiment of the present invention.

FIG. 2 is an exemplary diagram of a color factor system according to an embodiment of the present invention.

FIG. 3 is a diagram illustrating a color evaluation model according to an embodiment of the present invention.

FIG. 4 is a diagram illustrating an example of color quality prediction according to an embodiment of the present invention.

FIG. 5 is a diagram of an exemplary VR instructional resource color optimization model in an embodiment of the invention.

FIG. 6 is an exemplary diagram of an unobstructed VR teaching resource color scheme optimization in an embodiment of the invention.

FIG. 7 is a diagram illustrating an example color scheme for a VR instructional resource in an embodiment of the invention.

FIG. 8 is a Voronoi segmentation example diagram of VR instructional resource color patches in an embodiment of the invention.

FIG. 9 is a diagram illustrating an example of barrier-free modification of an original VR teaching resource color scheme in an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

As shown in fig. 1, the present embodiment provides a method for optimizing colors of barrier-free VR teaching resources for people with color vision anomaly, including the following steps:

(1) and constructing a color cognition factor system. Constructing a color factor system of the barrier-free VR teaching resource from the aspects of color rules, visual physiological structures and visual psychosocial feelings; respectively expressing the characteristics of the three types of color factors by adopting different parameters; the method comprises the steps of collecting color samples of VR scenes, constructing a color sample library, establishing an expert knowledge library of VR teaching resource color quality, designing various types of tasks, and analyzing color cognitive differences among different crowds.

(1-1) constructing a color factor system. As shown in fig. 2, color rule series factors are extracted from aspects of color matching, color discrimination, color significance and the like; and (3) constructing a physiological structure factor according to aspects such as the color blindness type, the color weakness type, the severity and the like, forming a color psychosensory factor according to aspects such as color harmony, color cooling and warming, color enthusiasm and the like, and constructing a color factor system of the barrier-free VR teaching resource.

(1-1-1) extracting the color rule series factors. And summarizing color design knowledge of VR teaching resources from aspects of color self-rules, color expression types, color semantics, object forms, distribution and the like, and extracting series factors such as color matching, distinguishing degree, uniformity, significance, internal semantic relation and the like.

(1-1-2) constructing physiological structure series factors of people with abnormal color vision. Analyzing the difference of the tested object in the visual physiological perception, such as the crowd with red-green achromatism, the red and green can be regarded as yellow, the purple can be regarded as blue, and establishing VR teaching resource color visual physiological structure series factors aiming at the difference of the color vision abnormal type (color blindness/color weakness), the color blindness type (red blindness, green blindness, etc.), the color weakness type (red weakness, green weakness, etc.), the color vision abnormal severity, and the like.

(1-1-3) extracting color psychology feeling series factors. Aiming at the overall visual perception of the VR teaching resource color, emotion series factors such as color harmony, color cooling and heating, color active/passive and the like are extracted, and the weight of psychological perception factors such as color harmony, color emotion and the like is judged according to the color quality degree influencing the barrier-free VR teaching resource.

And (1-2) characterizing the color factor. According to a CIEDE2000 color difference model, obtaining characteristics of perception difference, uniformity, change trend and the like of different colors; expressing visual physiological structure factors by using a color spectrum perception function or a color weakness spectrum perception curve; and calculating the difference values of lightness, chroma and hue to obtain color harmony characteristics, and calculating the color emotion characteristics by using the color temperature, lightness, positivity/negativity equivalence and hue angle.

(1-2-1) characterization expression of color rule series factors. Aiming at the problem that the RGB color model can not be directly converted into the CIELab color model, the RGB color space is firstly converted into the XYZ color space and then converted into the CIELab color space, and the conversion relation is shown in formulas (1) to (3).

In the CIELab color model, the lightness L has a value range of [0,100 ]]From pure black to pure white; the value of the a and b components is in the range of-128,127]Corresponding to the respective color transitions. X_N,Y_N,Z_NThe default coefficients are 95.047, 100.0, 108.883.

According to a CIEDE2000 color difference model, the difference values of three components of lightness, chroma and hue are used to obtain the perception difference characteristics among different colors, and the process is shown in a formula (4); calculating the distance between adjacent colors to obtain the uniformity characteristic of perception difference, wherein the process is shown in a formula (5); the variation trend characteristics of the color are extracted by using the monotonicity of the color, and the process is shown in formula (6).

The color difference perception difference characterization calculation formula is as follows:

the delta L, the delta C and the delta H respectively represent lightness difference, chroma difference and color difference of colors; k_L、K_C、K_HParameter factors corresponding to the three components of lightness, chroma and hue are correction coefficients related to experimental conditions; s_L、S_C、S_HThe weight functions are respectively corresponding to the lightness, chroma and hue components and are used for correcting the color space uniformity; r_TIs a rotation function to correct for the deflection of the color space blue region in the direction of the principal axis of the tolerance ellipse.

Color distance uniformity characterization calculation formula:

D_iis the color distance between adjacent colors in the color scheme,is the color distance mean.

The color monotonicity characterization calculation formula is as follows:

c₁、c₂the colors are numbered in sequence in the color scheme, and L is the color brightness.

(1-2-2) the characteristic expression of the physiological series factors of the people with abnormal color vision. Calculating a projection plane of the color blind population in the degraded color gamut by using the spectrum perception function of the color in the color blind population, wherein the process is as shown in a formula (7); calculating the interpolation position of the color on the color weakness spectrum perception curve, wherein the process is as the formula (8); thereby expressing visual physiological structural factors such as achromatopsia, color weakness and the like.

L, M, S is the responsivity of primary color at long, medium and short wavelengths in LMS color space, L_p、M_p、S_pFor red blind perception after conversionResponsivity of color, L_d、M_d、S_dThe responsivity of the color is perceived for the green blindness.

L (lambda), M (lambda) and S (lambda) are spectrum perception functions of normal people on long wave, medium wave and short wave, and L_a(λ)、M_a(λ)、S_a(lambda) is the spectral perception function of the population with color weakness, Delta lambda_L、Δλ_L、Δλ_LThe offset of the color-weak people on the spectrum perception curve is shown.

(1-2-3) characterization expression of color psychosensory series factors. Calculating the difference values of lightness, chroma and hue in a CIELab color space by using a formula (9) to obtain color harmony characteristics reflecting color visual psychology feelings; the color emotional characteristics are calculated by using the equivalent of color temperature, lightness, positivity and negativity and the hue angle in the color space, and the specific process is shown as a formula (10).

The color harmony characterization calculation formula is as follows:

C₁,C₂representing two different colors, Δ L_ab、ΔC_ab、ΔH_abLightness difference, chroma difference, hue difference, L in CIELab color space₁、L₂Lightness of two colors, CH (C)₁,C₂) Is a harmonic feature of two colors.

The color emotion characterization calculation formula is as follows:

WC color temperature, HL color weight, AP color active/passive value, C_abIs the color chroma, h_abFor colors in CIELab color spaceThe hue angles L, a and b are lightness, red-green chroma and yellow-cyan chroma of the colors in the space.

(1-3) color quality recognition. Selecting different color dimension indexes, and establishing an expert knowledge base of VR teaching resource color quality; designing various types of cognitive tasks around the theme matching of teaching resources; and (3) finishing different cognitive tasks by organizing the crowd with normal and abnormal color vision as a control group, and analyzing the color cognitive difference among different crowds.

(1-3-1) generating an expert knowledge base of color quality. And establishing an expert knowledge base of VR teaching resource color quality according to the dimensions of color discrimination, color uniformity, color semantics, color selection path, color monotonicity, color geometrical form, color space distribution, color harmony, color emotion and the like.

(1-3-2) designing a color cognition task. And designing various types of cognitive tasks such as positioning, recognition, comparison, sequencing, contact, recall and the like around factors such as hue difference, hue quantity, color distance, color area, color distribution, color internal semantic relation, color and teaching resource theme matching and the like.

(1-3-3) analyzing color recognition difference. By means of an Ali crowdsourcing platform, people with normal color vision and abnormal color vision as comparison groups are invited to complete corresponding color cognition tasks in a network anonymity mode, and color cognition differences among different people are analyzed by collecting learning behavior indexes of a tested object, such as accuracy, completion efficiency, eye movement indexes of watching duration, watching frequency, eye jump times and the like.

(2) And constructing a color evaluation model. Screening alternative factors of the color evaluation model, describing the characteristics of an input color scheme, and determining the weight of the factors; constructing color sample libraries of different disciplines, styles and types, generating color labels of the color samples, and extracting a color scheme of the teaching resources; designing a color evaluation regressor, and comprehensively evaluating the color quality prediction of the learner by adopting various indexes by using an Adaboost ensemble learning model.

(2-1) feature combinations of the color evaluation model. As shown in fig. 3, screening factors having significant influence on cognition of corresponding people as candidate factors of a color evaluation model; selecting alternative factors with high association degree by using a machine learning method, and describing the characteristics of the input color scheme; and determining the weight of the color evaluation model factor by applying a machine learning method.

(2-1-1) selecting alternative factors. According to the type of the color scheme and the cognition difference facing to the crowd, factors which have obvious influence on the cognition of the corresponding crowd are screened from three series factors of color rules, color vision physiological structures and color vision psychological feelings to serve as alternative factors of a color evaluation model.

(2-1-2) model characterization. According to the cognitive mechanism difference of different types of people on colors, from the dimensionalities of the VR teaching resource colors such as lightness, chroma, hue and harmony, a machine learning method is used for selecting alternative factors with high relevance and describing the characteristics of an input color scheme.

(2-1-3) determining the factor weight. According to the category (such as qualitative color, sequential color or two-end color) to which the VR teaching resource color scheme belongs, a machine learning method is applied to determine the influence weight of the characteristic factors of the color evaluation model, and the differences of the factors such as monotonicity, discrimination, uniformity, significance and harmony of the color scheme are highlighted.

And (2-2) extracting a color scheme. Collecting color samples of an existing VR scene, and constructing color sample libraries of different disciplines, styles and types; generating a color label of the sample based on the prediction result of the regressor and the actual category of the sample; and extracting the color characteristics of the picture by using the color histogram to obtain a complete color scheme.

(2-2-1) constructing a color sample library. Different scenes are collected from the existing VR games and VR teaching resources, and color samples are collected by utilizing hardware devices or software such as a spectrophotometry color meter, color acquisition software, a web crawler and the like to construct color sample libraries with different subjects, styles and types.

(2-2-2) color sample label generation. Aiming at the condition of color distribution imbalance in VR teaching resources, a GE-SMOTE method is selected to preprocess a data set, a DataBoost-IM method is combined to construct a mixing matrix, and a color label of a sample is generated based on the prediction result of a regressor and the actual category of the sample.

(2-2-3) color scheme extraction. Extracting the picture color features of the VR teaching resources by using a color histogram, setting a CIELab threshold value of each color, and tracking the boundary of the color in the image by adopting binarization processing to obtain a corresponding image spot area, as shown in FIG. 8; and obtaining a complete color scheme in the VR teaching resource picture through multiple iterations.

And (2-3) predicting color quality. As shown in fig. 4, a color evaluation regressor is designed, and a learner combination composed of a plurality of base learners is created by using a series connection, a parallel connection and the like; an Adaboost ensemble learning model is used, and the accuracy of the color sample training result is improved by combining a framework of a cascade learner; and comprehensively evaluating the color quality predicted by the learner by adopting various indexes to realize the color quality prediction of the VR teaching resources.

(2-3-1) color evaluation regressor design. Based on a CIELab color model, extracting color patches of VR teaching resources by using a k-means method, adopting classical regressor models such as KNN, logistic regression, decision tree and the like as base learners, and creating a learner combination consisting of a plurality of base learners by using modes such as series connection, parallel connection and the like.

(2-3-2) integrating learning model design. An Adaboost ensemble learning model is adopted, a framework of a cascade learner is combined, a historical judgment result and a current judgment result of a color scheme are comprehensively considered, an auxiliary judgment function is added, and the accuracy of a color sample training result is improved. The calculation steps of the Adaboost ensemble learning model are shown in formulas (11) to (18):

I. and initializing weight distribution of the training data. The samples in the training set are first given the same weight, i.e. each training sample acts the same in the base learner:

n is the total number of samples in the training set;

and II, iteratively training a base learner. The number of iterations is denoted by M, where M is 1,2,3, …, M

1.Using a weight distribution D_mLearning the training data set to obtain a basic classifier

G_m(x) X → { -1,1} equation (12)

2. Compute basis classifier G_m(x) Classification error rate e on training data set_m

Wherein ω is_m,iRepresents the weight of the ith sample in the mth round,I(G_m(x_i)≠y_i) For indicating the function, represent G_m(x_i)≠y_iWhen is, I (G)_m(x_i)≠y_i) 1, otherwise equal to 0.

3. Compute basis classifier G_mIs given by a weight coefficient alpha_m. The coefficient represents the basis classifier G_mThe weight occupied in the final classifier is calculated as follows:

e_mto classify the error rate, α_mWith e_mIs increased, i.e., the base classifier with smaller classification error takes up more weight in the final classifier.

4. And updating the weight distribution of the training data set for the next iteration.

Z_mIs a specification factor introduced to make D_m+1The probability distribution is specifically as follows:

repeating the steps 1 to 4 in the step II to obtain a series of weight parameters a_mAnd base classifier G_m。

Linearly combining the respective basis learners according to the weighting parameters.

With the sign () function, the continuous values of f (x) are converted to discrete values, so the final classifier is:

(2-3-3) color quality prediction. And comprehensively evaluating the color quality predicted by the learners by adopting indexes such as average absolute error, root mean square error, goodness of fit and the like, comparing the prediction performances of a plurality of learners, and selecting a strong learner with the optimal prediction performance to realize the color quality prediction of VR teaching resources for different crowds.

(3) And generating a color scheme optimization model. Supporting color evaluation generation facing different crowds, constructing a general optimal color selection scheme, and completing a customized color selection model facing specific color vision abnormal crowds; dividing the color pattern spots into different areas by using a Voronoi method, establishing a color selection model facing the cognition of people with abnormal color vision, and optimizing the design of a barrier-free VR teaching resource color scheme; screening out the optimal color scheme to generate a candidate generation scheme for optimizing the color selection; and establishing an optimized color selection model to realize barrier-free transformation of the original VR teaching resource color scheme.

(3-1) a color model for optimal selection of VR educational resource colors. As shown in fig. 5, different people are organized to score according to the litters scale, and color evaluation generation facing different people is supported; taking the color evaluation of the crowd with abnormal color vision as a constraint condition, and taking the color evaluation of the normal crowd as a target function to construct a general optimal color selection scheme; and (3) considering the cognitive characteristics of the specific color vision abnormal crowd, and constructing a customized color selection model.

(3-1-1) color evaluation generation. The method comprises the steps of associating colors of VR teaching resources with scenes with different geometric forms and different spatial layouts to form a series of VR teaching resources, organizing abnormal color vision and normal personnel to score according to a Likter scale, and supporting color evaluation generation facing different crowds.

(3-1-2) general most preferred color model. And constructing a color selection scheme of the general VR teaching resource by taking the color evaluation of the abnormal color vision crowd as a constraint condition and the color evaluation of the normal crowd as a target function, and solving the optimal color scheme relative to the normal crowd in the color series scheme meeting the cognition of the abnormal color vision crowd.

(3-1-3) customizing the color selection model. And (3) considering the cognitive characteristics of the specific color vision abnormal crowd, taking the color evaluation of the barrier-free VR teaching resources of the color vision abnormal crowd as a target function, constructing a customized color selection model, and solving a color scheme which maximally meets the cognition of the specific color vision abnormal crowd from a color space.

(3-2) constructing a barrier-free VR teaching resource color scheme, as shown in FIG. 6, and aiming at the color scheme of the existing sample, dividing a color pattern into different regions by using a Voronoi method; establishing a color selection scheme facing the cognition of people with abnormal color vision by using a color rule and an evaluation result as a constraint condition and a target function; and (3) introducing a firefly algorithm, and optimizing the construction of the barrier-free VR teaching resource color scheme.

(3-2-1) image segmentation. Aiming at a color sample library or a color scheme of an existing VR teaching resource (as shown in FIG. 7), the distribution of different colors in an image on a CIELab space is analyzed by utilizing kernel density, as shown in a public display (19), a spatial clustering method is adopted to merge similar color regions, the gravity center of a region polygon is calculated, a Voronoi diagram is used for segmenting the image, and a color pattern spot is divided into different regions, as shown in FIG. 8.

Kernel density estimation formula:

x_ithe ith data point in the sample is represented by K, the bandwidth is h, the Gaussian density function K (x) is the kernel function of kernel density estimation, and sigma is the sample variance.

(3-2-2) establishing a barrier-free color selection scheme. Combining the Voronoi subdivision-based image spots into different color schemes according to barrier-free color design rules, and establishing a barrier-free VR teaching resource color selection scheme facing the cognition of people with abnormal color vision by using the color rules and the evaluation results as constraint conditions and objective functions.

(3-2-3) optimizing the color scheme construction. And (3) introducing a firefly algorithm, generating an optimal solution by using a neural network algorithm in a color selection scheme, introducing the position of a candidate region in an optimal descrambling color scheme, improving a local search position updating formula, overcoming the defect of easy generation of a local optimal solution, and optimizing the construction of a barrier-free VR teaching resource color scheme.

The steps for optimizing the color scheme based on the firefly algorithm are shown in equations (20) to (21):

initializing firefly algorithm parameters: number of fireflies N, initial attraction degree beta₀Step size factor alpha, firefly initial position X₁(X_i1,X_i2,…,X_ik) And a maximum number of iterations T;

calculating the brightness of each firefly and sequencing: calculating the fitness f corresponding to each firefly_i(l_i,a_i,b_i) Corresponding to the brightness of the fireflies and sequencing to obtain the position of the fireflies with the maximum brightness;

m represents the order of various colors; i S_il_i-S_jl_jAnd | represents the moment difference between the colors of the color scheme, and if the sum of the absolute values of the moment differences of all the colors is smaller, the color matching is more harmonious. (l)₀,a₀,b₀) Lab values representing gray in CIELab color space;indicating the mixed color after mixing by area in the current color scheme, if closer to the reference color (l)₀,a₀,b₀) The higher the fitness of the color scheme, K₁、K₂、K₃Respectively representing the weights designed by the scoring expert.

Judging whether the iteration is finished: if the algorithm reaches the maximum iteration time T, the algorithm goes to the step IV, otherwise, the algorithm goes to the step V;

IV, outputting the position and the brightness of the firefly with the maximum brightness, and obtaining the position and the brightnessAs a color scheme;

v, updating the position of the firefly: the firefly location update rule is as follows

X_i＝X_i+β(r)×(X_j-X_i) + α × (rand-1/2) formula (21)

X_i，X_jRespectively representing the spatial positions of firefly i and firefly j;denotes the distance between firefly i and firefly j, where r_ij＝‖X_i-X_j‖；β₀Is represented by r_ijAttraction degree when 0; g represents a light intensity absorption coefficient; alpha represents a step-size factor, alpha is equal to 0,1](ii) a (rand-1/2) represents an interference term to avoid the algorithm falling into local optimality.

And (3-3) carrying out barrier-free modification on the original VR teaching resource color scheme. As shown in fig. 9, the optimal color scheme is selected according to the logic structure of the color; adopting Voronoi to divide color space and generating a potential area for optimizing color selection; and establishing an optimized color selection model taking the similarity as a constraint condition and the color evaluation difference as a target function, and realizing barrier-free reconstruction of the original VR teaching resource color scheme.

(3-3-1) selection of color scheme. Determining the category of a color scheme, such as qualitative color, sequential color, colors at two ends and the like, according to the logic structure of the original VR teaching resource color; and selecting a series of schemes which accord with corresponding rules in a color system or a color space by utilizing the design rules of the color schemes, and further screening out the optimal color scheme.

(3-3-2) generation of a color candidate optimization scheme. Based on a CIELab color space, taking the spatial distribution of the original VR teaching resource color scheme as a reference, adopting Voronoi to subdivide different color patches, wherein each generated Voronoi subspace is a candidate area for optimizing and selecting colors of the color scheme, and the color candidate optimization scheme can be formed by the combination of different areas.

(3-3-3) selection of an optimization scheme. Calculating the similarity of the VR teaching resource color scheme based on the color series factors, and judging the difference of the color style before and after optimization according to the contrast similarity; and establishing an optimized color selection model taking the similarity as a constraint condition and the color evaluation difference as a target function, and realizing barrier-free optimization of the original VR teaching resource color.

Details not described in the present specification belong to the prior art known to those skilled in the art.

It will be understood by those skilled in the art that the foregoing is merely a preferred embodiment of the present invention, and is not intended to limit the invention, such that any modification, equivalent replacement or improvement made within the spirit and principle of the present invention shall be included within the scope of the present invention.