1. A human body size calculation method based on a dense ellipse model comprises the following steps:

A. finding out common human body measurement auxiliary points on a three-dimensional SMPL template model;

B. selecting a size calculation point row according to the model measurement auxiliary points;

C. different size calculation methods are designed for different types of sizes.

2. The method of claim 1, wherein the finding of common body measurement auxiliary points on a three-dimensional SMPL template model in section a is: marking 30 measurement auxiliary points required by measurement on the three-dimensional human body model according to the relevant regulation of national standards on the reference points, and determining the vertex serial number, the three-dimensional coordinates and the coordinate system of the measurement auxiliary points on the model by using MeshLab software.

3. The method according to claim 1, wherein in part B, the selecting the row of size calculation points according to the model measurement auxiliary points is: and according to the auxiliary measuring points, continuously using MeshLab software to select a measuring point column with high correlation corresponding to the sizes of all parts of the body.

4. The method of claim 1, wherein the step of designing different sizes for different types of sizes in the section C comprises: for the length of the straight line segment, the coordinate difference of a plurality of measurement characteristic points in a certain direction or the length of a vector is calculated only according to the human body measurement characteristic point column. For the special arc length dimension of the back and waist length class, the special arc length dimension is regarded as being formed by small segments of arc lengths, the length of each segment of arc is approximate to the length of an internal chord of the segment of arc, and the whole arc length can be regarded as being formed by overlapping a plurality of internal chord lengths. For the length of a closed curve segment, a dense ellipse model algorithm is provided, measuring points corresponding to the closed circumference size are projected towards a certain direction to obtain a corresponding two-dimensional curve, the two-dimensional curve can be regarded as being composed of ellipse segments with different curvatures, and the calculated size is obtained by accumulating the ellipse curve segments.

Background

The traditional human body measurement mainly adopts contact measurement, and measurement personnel with abundant experience measures the datum points and datum lines of the human body by adopting a distance meter, a human body measuring instrument, a height meter, a tape measure and the like face to face according to the position, the size and the body proportion of each part of the human body to obtain the surface dimensions of each part of the human body, such as length or girth and the like. The methods are intuitive and simple to use tools, but they do not meet the need for remote acquisition of dimensional information and are affected by many other factors. This is still a challenging task due to the ambiguity of depth and scale, complex background and human body structure, and in addition, for people of special body type, it is too fat or too thin, the feature point extraction is not accurate, resulting in large size calculation error.

In recent years, in order to obtain the size of a human body, a 3D three-dimensional human body model based on 2D image reconstruction becomes one of the directions of human body measurement research, and a common method adopts a method of selecting characteristic point rows related to all parts of a human body on the model for human body size measurement and calculating the size of the human body by using a spatial relationship between the characteristic point rows. However, when the closed circumference size is calculated, the method only utilizes length accumulation between point columns or uses single ellipse or double ellipse fitting to calculate, and does not consider that the surface of the human body is an irregular curved surface, so that the size calculation error is large. This method is thus improved by proposing a dense elliptical model, considering the closed circumference as being formed by stitching a plurality of elliptical arc lengths of different curvatures.

Disclosure of Invention

In order to solve the problem of large fitting error of the closed circumference size, the invention provides a dense ellipse model, the closed circumference size is considered to be formed by sewing a plurality of quarter ellipses with different curvatures, and the length of a corresponding ellipse section is calculated through an included angle formed by a point row selected from the circumference size and the circle center.

The technical scheme adopted by the invention is as follows:

a human body size calculation method based on a dense ellipse model comprises the following steps:

A. finding out common human body measurement auxiliary points on a three-dimensional SMPL template model;

B. selecting a size calculation point row according to the model measurement auxiliary points;

C. designing different size calculation methods for different types of sizes;

in part a, the commonly used anthropometric auxiliary points are found on a three-dimensional template model: marking 30 measurement auxiliary points required by measurement on the three-dimensional human body model according to the relevant regulation of national standards on the reference points, and determining the vertex serial number, the three-dimensional coordinates and the coordinate system of the measurement auxiliary points on the model by using MeshLab software.

In part B, the selecting the size calculation point sequence according to the model measurement auxiliary points means: and according to the auxiliary measuring points, continuously using the MeshLab to select a measuring point column which forms the corresponding size of each part of the body and has high correlation.

In part C, the dimension calculation method for designing different dimensions for different types refers to: for the length of the straight line segment, the coordinate difference of a plurality of measurement characteristic points in a certain direction or the length of a vector is calculated only according to the human body measurement characteristic point column. For the special arc length dimension of the back and waist length class, the special arc length dimension is regarded as being formed by small segments of arc lengths, the length of each segment of arc is approximate to the length of an internal chord of the segment of arc, and the whole arc length can be regarded as being formed by overlapping a plurality of internal chord lengths. For the length of the closed curve segment, the traditional method estimates through an ellipse model, however, because the size of the human body is formed by irregular curve segments, the size error of different areas of the human body estimated by using a single ellipse or a double ellipse is larger. Therefore, we propose a dense ellipse model under polar coordinates, and project the measurement points corresponding to the enclosed dimension of circumference towards a certain direction to obtain a corresponding two-dimensional curve, which can be regarded as being composed of ellipse segments with different curvatures, and the calculated dimension is obtained by the accumulation of the ellipse curve segments, as shown in fig. 4.

Knowing the centre point O and the coordinates of the rows of points constituting the closed curve, a_iAnd a_i+1Is the boundary point of the ellipse segment, Oa_iAnd Oa_i+1Is the length of the edge, the angle theta can be calculated from the line segment, the point P can be obtained by making a perpendicular line from the end point of the shorter side to the longer side, and OP, Pa can be obtained_i、Pa_i+1Length of (1), with P as the origin of coordinates and Pa_iAnd Pa_i+1The middle and longer sides are used as a long semi-axis L, the shorter side is used as a short semi-axis s to establish an elliptic equation L, and the arc length E_iFor a quarter ellipse perimeter, the equations define (1), (2) as follows:

adding the ellipse sections with different curvatures to obtain a closed circumference size, wherein the formula is defined as (3), wherein E_allRepresenting the dimension to be measured, N representing the corresponding angle of each elliptical segment, E_NIs represented by a_Na₁The length of the oval segment formed.

Drawings

In order to more clearly explain the technical solution of the present invention, the drawings needed to be used in the summary of the invention will be briefly described below.

FIG. 1 is a design flow chart of a method for obtaining human body dimensions based on a three-dimensional human body model according to the present invention.

FIG. 2 is a schematic diagram of a commonly used reference point for measuring a human body marked by the method.

FIG. 3 is a schematic diagram of the dimensions and symbolic illustration of the three-dimensional model labeled by the method.

FIG. 4 is a schematic diagram of the dense ellipse model algorithm labeled by the method.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in further detail below.

The standard of this embodiment is that 100 groups of human body front, side and back images and 38 pieces of real size information to be measured are collected in advance. The experimental equipment is a Linux system desktop, an Inviada 2080TiGPU and an Intel i7 CPU.

The designed algorithm structure is realized by programming of a python language and a pytorech framework.

The accuracy rate of size calculation based on the three-dimensional human body model without the method is 96%, and the accuracy rate reaches 97.6% after the method is adopted.