1. A method for searching a picture by a picture is characterized by comprising the following steps:

acquiring an image to be searched;

respectively calculating the similarity between the image to be searched and the standard image of each category group in a preset image library; the preset image library is divided into a plurality of category groups, and each category group comprises a standard image and other images except the standard image;

determining a plurality of target category groups meeting preset conditions according to the similarity;

obtaining the number sequence of the images corresponding to the target category group according to the target category group and a database; the serial number of the image, the category group where the image is located and the similarity between the image and the standard graph of the category group are stored in the database as records, and the sequencing sequence of the serial number of each image in the serial number sequence is determined based on the similarity between each image and the standard graph of the category group;

and reading corresponding images from the preset image library in sequence as image search results according to the number sequence of the images.

2. The method of claim 1, wherein the updating process of the preset gallery comprises:

respectively calculating the similarity between the image to be added and the standard images of each category group in the preset image library, and comparing the similarity with a preset similarity threshold value:

if the similarity is larger than or equal to the similarity threshold, taking the category group corresponding to the maximum value of the similarity as the category group where the image to be added is located;

otherwise, creating a category group, storing the image to be added into a buffer area, using the image to be added as a standard diagram of the created category group, and using the created category group as the category group where the image to be added is located.

3. The method for searching a graph according to claim 2, further comprising:

and storing the serial number of the image to be added, the category group of the image to be added and the corresponding similarity in the database as records.

4. The method for searching the image according to the claim 1 or 3, wherein the calculating the similarity between the image to be searched and the standard image of each category group in the preset image library respectively comprises:

performing feature analysis on the image according to an image recognition algorithm, extracting a feature value of the image, and respectively comparing the feature value of the image with the feature value of a standard image of each category group in a preset image library to obtain similarity; and the characteristic value of the standard graph of each category group is obtained by performing characteristic analysis on the standard graph according to an image recognition algorithm and extracting the characteristic value.

5. The method as claimed in claim 1, wherein determining a plurality of target category groups satisfying a predetermined condition according to the similarity comprises:

and performing descending order arrangement on the category groups according to the similarity, and selecting the previous category groups as target category groups.

6. The method for searching in a graph according to claim 1, wherein obtaining a number sequence of corresponding images according to the target category group and the database comprises:

searching records in the database according to the target category group to obtain the number and the similarity of the corresponding images of the target category group;

and sorting the records in a descending order according to the similarity to obtain a number sequence of the corresponding image.

7. A method as claimed in claim 1, wherein the similarity threshold is 80%.

8. A system for searching graphs using the method of claims 1-7, comprising:

the image acquisition module is used for acquiring an image to be searched;

the similarity calculation module is used for calculating the similarity between the image to be searched and the standard images of each category group in the preset image library respectively; the preset image library is divided into a plurality of category groups, and each category group comprises a standard image and other images except the standard image;

the target category group determination module is used for determining a plurality of target category groups meeting preset conditions according to the similarity;

the number sequence determining module is used for obtaining the number sequence of the images corresponding to the target category group according to the target category group and the database; the serial number of the image, the category group where the image is located and the similarity between the image and the standard graph of the category group are stored in the database as records, and the sequencing sequence of the serial number of each image in the serial number sequence is determined based on the similarity between each image and the standard graph of the category group;

and the image reading module is used for sequentially reading the corresponding images from the preset image library as image searching results according to the number sequence of the images.

9. The system for searching for images according to claim 8, wherein the calculating the similarity between the image to be searched and the standard images of each category group in the preset image library respectively comprises:

performing feature analysis on the image according to an image recognition algorithm, extracting a feature value of the image, and respectively comparing the feature value of the image with the feature value of a standard image of each category group in a preset image library to obtain similarity; and the characteristic value of the standard graph of each category group is obtained by performing characteristic analysis on the standard graph according to an image recognition algorithm and extracting the characteristic value.

10. The system for searching for a graph according to claim 8, wherein determining a plurality of target category groups satisfying a predetermined condition according to the similarity comprises:

and performing descending order arrangement on the category groups according to the similarity, and selecting the previous category groups as target category groups.

Background

With the increasing of image data information in the internet, the demand of users for image search is also increasing, the search mode of image search is applied more and more widely, and users can search similar images in a gallery by inputting images.

The method for searching the images by the images is to input the images to be searched into a gallery, and retrieve the images which are the same as or similar to the images to be searched in the gallery. In the prior art, a method for searching a picture needs to compare the similarity between an image to be searched and each image in a picture library, and select an image with high similarity as a target image. When the image data in the gallery is more, the problems of more calculation resources and longer search time exist.

At present, the image searching method generally compares the similarity of an image to be searched with each image in a gallery in sequence, and selects an image with higher similarity as a target image. When the number of images in the gallery reaches tens of thousands or more, a traditional method for searching and searching images for a target image needs to consume a large amount of computing resources (generally, a picture processing server configured with a GPU needs to be adopted), so that the problems of long search time and low search efficiency exist, and the requirements of users cannot be met.

Disclosure of Invention

The invention aims to provide a method and a system for searching a picture by using a picture, which aim to solve the technical problems of more calculation resources and longer search time in the conventional method for searching the picture by using the picture.

The purpose of the invention can be realized by the following technical scheme:

acquiring an image to be searched;

respectively calculating the similarity between the image to be searched and the standard image of each category group in a preset image library; the preset image library is divided into a plurality of category groups, and each category group comprises a standard image and other images except the standard image;

determining a plurality of target category groups meeting preset conditions according to the similarity;

obtaining the number sequence of the images corresponding to the target category group according to the target category group and a database; the serial number of the image, the category group where the image is located and the similarity between the image and the standard graph of the category group are stored in the database as records, and the sequencing sequence of the serial number of each image in the serial number sequence is determined based on the similarity between each image and the standard graph of the category group;

and reading corresponding images from the preset image library in sequence as image search results according to the number sequence of the images.

Optionally, the updating process of the preset gallery includes:

respectively calculating the similarity between the image to be added and the standard images of each category group in the preset image library, and comparing the similarity with a preset similarity threshold value:

if the similarity is larger than or equal to the similarity threshold, taking the category group corresponding to the maximum value of the similarity as the category group where the image to be added is located;

otherwise, creating a category group, storing the image to be added into a buffer area, using the image to be added as a standard diagram of the created category group, and using the created category group as the category group where the image to be added is located.

Optionally, the method further comprises:

and storing the serial number of the image to be added, the category group of the image to be added and the corresponding similarity in the database as records.

Optionally, the calculating the similarity between the image to be searched and the standard graph of each category group in the preset gallery respectively includes:

performing feature analysis on the image according to an image recognition algorithm, extracting a feature value of the image, and respectively comparing the feature value of the image with the feature value of a standard image of each category group in a preset image library to obtain similarity; and the characteristic value of the standard graph of each category group is obtained by performing characteristic analysis on the standard graph according to an image recognition algorithm and extracting the characteristic value.

Optionally, determining, according to the similarity, a plurality of target category groups that satisfy a preset condition includes:

and performing descending order arrangement on the category groups according to the similarity, and selecting the previous category groups as target category groups.

Optionally, obtaining a number sequence of corresponding images according to the target category group and the database includes:

searching records in the database according to the target category group to obtain the number and the similarity of the corresponding images of the target category group;

and sorting the records in a descending order according to the similarity to obtain a number sequence of the corresponding image.

Optionally, the similarity threshold is 80%.

The invention also provides a system for searching the picture by the picture, which comprises:

the image acquisition module is used for acquiring an image to be searched;

the similarity calculation module is used for calculating the similarity between the image to be searched and the standard images of each category group in the preset image library respectively; the preset image library is divided into a plurality of category groups, and each category group comprises a standard image and other images except the standard image;

the target category group determination module is used for determining a plurality of target category groups meeting preset conditions according to the similarity;

the number sequence determining module is used for obtaining the number sequence of the images corresponding to the target category group according to the target category group and the database; the serial number of the image, the category group where the image is located and the similarity between the image and the standard graph of the category group are stored in the database as records, and the sequencing sequence of the serial number of each image in the serial number sequence is determined based on the similarity between each image and the standard graph of the category group;

and the image reading module is used for sequentially reading the corresponding images from the preset image library as image searching results according to the number sequence of the images.

Optionally, the calculating the similarity between the image to be searched and the standard graph of each category group in the preset gallery respectively includes:

performing feature analysis on the image according to an image recognition algorithm, extracting a feature value of the image, and respectively comparing the feature value of the image with the feature value of a standard image of each category group in a preset image library to obtain similarity; and the characteristic value of the standard graph of each category group is obtained by performing characteristic analysis on the standard graph according to an image recognition algorithm and extracting the characteristic value.

Optionally, determining, according to the similarity, a plurality of target category groups that satisfy a preset condition includes:

and performing descending order arrangement on the category groups according to the similarity, and selecting the previous category groups as target category groups.

The invention provides a method and a system for searching a picture by a picture, wherein the method comprises the following steps: acquiring an image to be searched; respectively calculating the similarity between the image to be searched and the standard image of each category group in a preset image library; the preset image library is divided into a plurality of category groups, and each category group comprises a standard image and other images except the standard image; determining a plurality of target category groups meeting preset conditions according to the similarity; obtaining the number sequence of the images corresponding to the target category group according to the target category group and a database; the serial number of the image, the category group where the image is located and the similarity between the image and the standard graph of the category group are stored in the database as records, and the sequencing sequence of the serial number of each image in the serial number sequence is determined based on the similarity between each image and the standard graph of the category group; and reading corresponding images from the preset image library in sequence as image search results according to the number sequence of the images.

In view of the above, the invention brings the following beneficial effects:

the preset image library is divided into a plurality of category groups, when the image to be searched is searched in the preset image library according to the image to be searched, similarity comparison between the image to be searched and each image in the image library is not required in sequence, the similarity comparison between the image to be searched and a standard image of each category group is only required, the comparison frequency is the number of the category groups in the preset image library, the calculation resources required to be consumed are reduced, the time for searching the image is shortened, and the image searching efficiency is improved; the invention classifies and manages the images in the large-capacity gallery in batches through the category group, and can realize the quick search of the large-capacity gallery by using low-calculation-force resources when searching the images by using the images.

Drawings

FIG. 1 is a schematic flow chart illustrating a method for searching a graph according to the present invention;

FIG. 2 is a schematic diagram of a system for searching a graph according to the present invention.

Detailed Description

The embodiment of the invention provides a method and a system for searching a picture by using a picture, which aim to solve the technical problems of more calculation resources and longer search time in the conventional method for searching the picture by using the picture.

To facilitate an understanding of the invention, the invention will now be described more fully with reference to the accompanying drawings. Preferred embodiments of the present invention are shown in the drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.

Referring to fig. 1, the following is an embodiment of a method for searching a graph according to the present invention, including:

s100: acquiring an image to be searched;

s200: respectively calculating the similarity between the image to be searched and the standard image of each category group in a preset image library; the preset image library is divided into a plurality of category groups, and each category group comprises a standard image and other images except the standard image;

s300: determining a plurality of target category groups meeting preset conditions according to the similarity;

s400: obtaining the number sequence of the images corresponding to the target category group according to the target category group and a database; the serial number of the image, the category group where the image is located and the similarity between the image and the standard graph of the category group are stored in the database as records, and the sequencing sequence of the serial number of each image in the serial number sequence is determined based on the similarity between each image and the standard graph of the category group;

s500: and reading corresponding images from the preset image library in sequence as image search results according to the number sequence of the images.

In step S100, an image to be searched input by a user is obtained, and the user may select an image from a network or a local area as the image to be searched, or may select an image in a preset gallery as the image to be searched.

In this embodiment, before searching the images with the images, the preset gallery needs to be divided into a plurality of category groups, and each category group includes one standard image and other images except the standard image. Adding the images into a preset image library, performing feature analysis on each image in the preset image library through polling processing by using idle time computing power resources, extracting a feature value of each image, respectively calculating the similarity between the images and the standard images of each category group in the preset image library, and comparing the similarity with a preset similarity threshold value:

(1) when the similarity is greater than or equal to a preset similarity threshold, taking a category group corresponding to the maximum value of the similarity as a category group in which the image is positioned; meanwhile, the serial number of the image, the category group of the image and the corresponding similarity are taken as records and stored in a database, but the image is not stored in an SDK or a cache area of a server;

(2) when the similarity is smaller than a preset similarity threshold, a category group is newly established, the image is used as a standard graph of the category group, the newly established category group is used as the category group where the image is located, and the image is stored in an SDK or a buffer area of a server; at the same time, the number of the image, the category group in which the image is located, and the corresponding similarity are stored in the database as records.

It should be noted that, since the similarity in recording is the similarity between another image (an image other than the standard chart in the same category group) and the standard chart, when a certain image is used as the standard chart of the new category group, the similarity value of the image corresponding to the recording is 100%.

The following describes a process of dividing the gallery into a plurality of category groups with reference to a specific example:

firstly, when only one image in the image library is image 1, the image 1 is subjected to feature analysis by using an image recognition algorithm and a feature value T of the image 1 is extracted₁At this time, there is no category group in the gallery, so a category group faceID needs to be created₁Image 1 as a class group faceID₁Standard graph of (1), FaceID₁As the category group of the image 1, storing the number of the image, the category group of the image and the corresponding similarity as records in a database;

it is worth noting that each category group in the gallery has only 1 standard map, and once the standard maps for a category group are determined, they do not change, and there is only one standard map in the category group FaceID1, image 1.

② when there are two images in the gallery, image 1 and image 2, firstly, divide image 1 into class group faceID₁Then, the image 2 needs to be divided into classification groups; carrying out feature analysis on the image 2 by using an image recognition algorithm and extracting a feature value T of the image₂Will T₂FaceID with class group₁Comparing the characteristic values of the standard graph to obtain the similarity S₁：

If S₁Not less than the similarity threshold (e.g. 80%), therefore, there are only 1 category group in the temporal atlas, and there is no need to continue the similarity comparison, i.e. the category group in which the image 2 is located is faceID₁The number of the image 2 and the category group faceID of the image 2₁Similarity S₁Storing the data in a local database;

if S₁If the similarity threshold is less than 80%, a category group FaceID is newly established₂Image 2 as a class group faceID₂The category group of the image 2 is faceID₂(ii) a It is worth noting that the similarity S is now set₁The valuation is 100%; the number of the image 2 and the FaceID of the category group₂Similarity S₁And storing the data in a local database.

When three images, namely image 1, image 2 and image 3, exist in the gallery, the image 1 and the image 2 are divided into corresponding classification groups, and then the image 3 needs to be divided into the classification groups. There are two possible cases at this time:

case 1, image 1 and image 2 belong to the same class group faceID₁I.e. only 1 category group faceID in the current gallery₁；

Case 2, image 1 and image 2 belong to the category group faceID, respectively₁And faceID₂I.e. there are 2 class groups faceID in the current gallery₁And faceID₂；

For case 1, adopt method of 2, image 3 and class group faceID₁Comparing the standard graphs to obtain corresponding similarity and judging;

for case 2, image 3 is associated with a class group faceID, respectively₁、FaceID₂The standard image is compared with the similarity, specifically, the image 3 is subjected to characteristic analysis by using an image recognition algorithm and a characteristic value T of the image is extracted₃Will T₃faceID with class group, respectively₁Characteristic value T of the standard map₁、FaceID₂Characteristic value T of the standard map₂Comparing to obtain corresponding similarity S₁And S₂：

If S₁A similarity threshold (e.g. 80%) and S₂The similarity threshold (such as 80%) is equal to or more than S₁And S₂The corresponding category group is taken as the category group in which the image 3 is located; that is, when S₁>S₂When the image 3 is in the category group, the category group is faceID₁Otherwise, the category group of the image 3 is faceID₂；

If S₁A similarity threshold (e.g. 80%) and S₂<If the similarity threshold is 80%, the category group of the image 3 is faceID₁；

If S₁<Similarity threshold (e.g. 80%) and S₂The similarity threshold is more than or equal to (for example, 80%), then the category group in which the image 3 is located is faceID₂；

If S₁<Similarity threshold (e.g. 80%) and S₂<If the similarity threshold is 80%, a new category group FaceID is created₃Image 3 as a class group faceID₃The category group of the image 3 is faceID₃。

By analogy, according to this method, the category group in which each image in the gallery is located can be determined, assuming that the images in the gallery are divided into K category groups.

It should be noted that each image added to the gallery has its corresponding category group, and the number, category group, and corresponding similarity of each image are stored as records in the local database.

In step S200, calculating similarity between the image to be searched and the standard image of each category group in the preset gallery respectively; the preset image library is divided into a plurality of category groups, and each category group comprises a standard image and other images except the standard image.

In this embodiment, an image to be searched input by a user is obtained, at this time, there are K category groups in the gallery, each category group has one standard graph, and each image in the gallery belongs to one of the category groups. Respectively calculating the similarity between the image to be searched and the standard image of each category group in the preset image library, specifically: and performing characteristic analysis on the image to be searched according to an image recognition algorithm, extracting a characteristic value of the image, and comparing the characteristic value with the characteristic value of the standard graph corresponding to each category group to obtain corresponding K similarity.

It is worth to be noted that, when a category group is newly created, the standard graph of the category group is stored in the SDK or the server buffer area, and the characteristic value of the standard graph of the category group is stored; and the characteristic values of the standard graphs of the category groups are obtained by performing characteristic analysis and extraction according to an image recognition algorithm.

S300: and determining a plurality of target category groups meeting preset conditions according to the similarity.

In step S300, the K similarity values obtained may be sorted in a descending order according to the similarity values, and the TOP M class groups with greater similarity values are selected as the target class groups by using the TOP N mechanism; wherein M is a natural number and is more than or equal to 1 and less than or equal to K.

It is understood that the user can also set a query threshold value when searching the graph to control the number of the selected target category groups according to actual needs. For example, a query threshold is set to 50, which means that the number of selected target category groups is at most 50, and when the number of target category groups satisfying the preset condition in the gallery is less than 50, the actual number of target category groups is selected, that is, the category groups are selected as the target category groups as long as how many category groups satisfy the preset condition.

S400: obtaining the number sequence of the images corresponding to the target category group according to the target category group and a database; the serial number of the image, the category group where the image is located and the similarity between the image and the standard graph of the category group are stored in the database as records, and the sequencing sequence of the serial number of each image in the serial number sequence is determined based on the similarity between each image and the standard graph of the category group.

In step S400, the M target class groups obtained according to the TOP N mechanism are arranged in descending order according to the similarity between the image to be searched and the standard images of the class groups; and for the images in the same target category group, performing descending order according to the similarity between other images (images except the standard images in the same category group) and the standard images of the category group to obtain the number sequences of the images corresponding to the M target category groups.

S500: and reading corresponding images from the preset image library in sequence as image search results according to the number sequence of the images.

In this embodiment, when searching for images by using images, the image to be searched input by the user only needs to be matched with the faceID of each category group_iAnd (i is more than or equal to 1 and less than or equal to K), returning the first M category groups with larger similarity as target category groups, searching records corresponding to the target category groups in the database according to the target category groups, performing descending order on the records according to the similarity in the records, and selecting the number of the corresponding image from the ordered records to obtain the number sequence of the images corresponding to the M target category groups. In this embodiment, by reasonably limiting the number of category groups, even tens of thousands of images can be displayed by using hundreds of category groupsAnd the method can save a large amount of computing resources and realize the quick search of the large-capacity gallery by using the small computing resources.

The method for searching for a picture by using a picture provided by the embodiment sufficiently utilizes computing power resources when a server is idle to divide images in a picture library into a plurality of category groups, each category group comprises a standard picture, one category group comprises one or more images, and when the picture is searched for in a preset picture library according to the images to be searched, similarity comparison between the images to be searched and each image in the picture library is not required, the similarity comparison between the images to be searched and the standard pictures of each category group is only required, the comparison times are only the number of the category groups in the preset picture library, so that the computing power resources required to be consumed are reduced, the time for searching the images is shortened, and the efficiency for searching the images is improved. According to the method and the device, the images in the large-capacity gallery are classified and managed in batch through the classification group, and when the images are searched in the map, the large-capacity gallery can be quickly searched by using low-calculation-force resources.

In the embodiment, the images in the large-capacity gallery can be classified and managed in batch through the category groups, the problem that each image occupies the gallery capacity in the cache region of the server is avoided, the problem of algorithm gallery capacity limitation is relieved, idle calculation resources of the algorithm server can be fully utilized, and longitudinal load balance in each time period is realized.

Referring to fig. 2, the present invention further provides an embodiment of a system for searching a graph with a graph, including:

the image acquisition module is used for acquiring an image to be searched;

the similarity calculation module is used for calculating the similarity between the image to be searched and the standard images of each category group in the preset image library respectively; the preset image library is divided into a plurality of category groups, and each category group comprises a standard image and other images except the standard image;

the target category group determination module is used for determining a plurality of target category groups meeting preset conditions according to the similarity;

the number sequence determining module is used for obtaining the number sequence of the images corresponding to the target category group according to the target category group and the database; the serial number of the image, the category group where the image is located and the similarity between the image and the standard graph of the category group are stored in the database as records, and the sequencing sequence of the serial number of each image in the serial number sequence is determined based on the similarity between each image and the standard graph of the category group;

and the image reading module is used for sequentially reading the corresponding images from the preset image library as image searching results according to the number sequence of the images.

In the embodiment, an image to be searched is acquired through an image acquisition module, the similarity between the image to be searched and a standard image of each category group in a preset image library is respectively calculated through a similarity calculation module, a plurality of target category groups are acquired through a target category group determination module, a number sequence of a corresponding image is obtained by searching in a database through a number sequence determination module, and a corresponding image is read from the image library as an image search result through an image reading module.

The preset image library in the embodiment is divided into a plurality of category groups, when the image to be searched is searched in the preset image library according to the image to be searched, similarity comparison between the image to be searched and each image in the image library is not required in sequence, the similarity comparison between the image to be searched and a standard image of each category group is only required, the comparison times are the number of the category groups in the preset image library, the calculation resources required to be consumed are reduced, the time for searching the image is shortened, and the image searching efficiency is improved; the invention classifies and manages the images in the large-capacity gallery in batches through the category group, and can realize the quick search of the large-capacity gallery by using low-calculation-force resources when searching the images by using the images.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.