CN111476253B

CN111476253B - Clothing image classification method, device and equipment and image classification method and device

Info

Publication number: CN111476253B
Application number: CN201910063759.7A
Authority: CN
Inventors: 赵永飞; 龙一民; 袁炜; 徐博文; 吴剑; 胡露露; 张民英; 神克乐; 刘志敏; 陈新; 尹宁
Original assignee: Alibaba Group Holding Ltd
Current assignee: Alibaba Group Holding Ltd
Priority date: 2019-01-23
Filing date: 2019-01-23
Publication date: 2024-04-02
Anticipated expiration: 2039-01-23
Also published as: CN111476253A; WO2020151529A1; TW202109356A

Abstract

The embodiment of the application provides a clothing image classification method, an image classification method, a device and equipment, wherein the clothing image classification method comprises the steps of acquiring a plurality of clothing images; dividing a garment image into a plurality of area images for any one garment image; respectively extracting color histogram features of the plurality of region images; weighting the color histogram features of the plurality of area images to obtain the image features of the clothing image; clustering the plurality of clothing images based on the image features of the plurality of clothing images to obtain a plurality of image sets; the number of images of the plurality of image sets corresponding to at least one garment image, respectively, is determined. The technical scheme provided by the embodiment of the application improves the classification accuracy of the clothing.

Description

Clothing image classification method, device and equipment and image classification method and device

Technical Field

The embodiment of the application relates to the technical field of computer application, in particular to a clothing image classification method, device and equipment.

Background

At present, in order to facilitate management and improve working efficiency, an existing clothing processing factory only processes a fixed work by using one production unit, so as to realize the production process of batch clothing, namely, a production line working mode in a common sense.

Since a clothing processing factory may process different clothing at the same time, in order to obtain the number of different clothing to be processed, statistics on the number of different clothing to be processed are required. The method is characterized in that a camera is generally used for collecting images of the clothing, the sift characteristic points, textures, edge contours and other characteristics of the images are extracted to serve as image characteristics, and then the images can be clustered based on the image characteristics to obtain images belonging to different types, so that the processing condition of the clothing corresponding to a certain type of image is confirmed.

However, since the types of garments are diversified, the acquired garment images may include garment images having relatively abundant textures, for example, images acquired for garments having relatively abundant patterns, or garment images having relatively simple textures, for example, images acquired for garments having pure colors, and the characteristics of the garment images, such as textures and contours, are too complex or too simple, and when the garment images are classified using the characteristics of the garment images, such as textures, the result is inaccurate, and classification errors occur.

Disclosure of Invention

The embodiment of the application provides a clothing image classification method, device and equipment, which are used for solving the technical problem that a plurality of image classification results are not accurate enough in the prior art.

In a first aspect, an embodiment of the present application provides a clothing image classification method, including:

acquiring a plurality of clothing images;

dividing a garment image into a plurality of area images for any one garment image;

respectively extracting color histogram features of the plurality of region images;

weighting the color histogram features of the plurality of area images to obtain the image features of the clothing image;

clustering the plurality of clothing images based on the image features of the plurality of clothing images to obtain a plurality of image sets;

determining the number of images of the plurality of image sets corresponding to at least one garment image respectively

In a second aspect, an embodiment of the present application provides an image classification method, including:

acquiring a plurality of images to be processed;

dividing an image to be processed into a plurality of area images aiming at any image to be processed;

weighting the color histogram features of the plurality of region images to obtain image features of the image to be processed;

classifying the images to be processed based on the image characteristics of the images to be processed to obtain a plurality of image sets;

Determining the image type of at least one image to be processed, which corresponds to each of the plurality of image sets;

and determining the number of images of at least one image to be processed corresponding to each image category.

In a third aspect, an embodiment of the present application provides a garment image classification apparatus, including:

the first acquisition module is used for acquiring a plurality of clothing images;

a first dividing module for dividing a clothing image into a plurality of area images for any one clothing image;

the first extraction module is used for respectively extracting the color histogram features of the plurality of area images;

the first weighting module is used for carrying out weighting processing on the color histogram features of the plurality of regional images to obtain image features of the clothing image;

the first clustering module is used for clustering the plurality of clothing images based on the image characteristics of the plurality of clothing images to obtain a plurality of image sets;

and the first determining module is used for determining the number of images of the plurality of image sets, which respectively correspond to at least one clothing image.

In a fourth aspect, an embodiment of the present application provides an image classification apparatus, including:

the second acquisition module is used for acquiring a plurality of images to be processed;

The second dividing module is used for dividing any image to be processed into a plurality of area images;

the second extraction module is used for respectively extracting the color histogram features of the plurality of area images;

the second weighting module is used for carrying out weighting processing on the color histogram characteristics of the plurality of regional images to obtain image characteristics of the image to be processed;

the second clustering module is used for classifying the plurality of images to be processed based on the image characteristics of the plurality of images to be processed to obtain a plurality of image sets;

the second determining module is used for determining the image type of at least one image to be processed, which corresponds to the plurality of image sets respectively;

and the third determining module is used for determining the image quantity of at least one image to be processed corresponding to each image category.

In a fifth aspect, in an embodiment of the present application, there is provided a garment image classification apparatus, including: the device comprises a storage component and a processing component, wherein the storage component stores one or more computer instructions which are used for the processing component to call and execute;

the processing assembly is configured to:

acquiring a plurality of clothing images; dividing a garment image into a plurality of area images for any one garment image; respectively extracting color histogram features of the plurality of region images; weighting the color histogram features of the plurality of area images to obtain the image features of the clothing image; clustering the plurality of clothing images based on the image features of the plurality of clothing images to obtain a plurality of image sets; the number of images of the plurality of image sets corresponding to at least one garment image, respectively, is determined.

In a sixth aspect, in an embodiment of the present application, there is provided an image classification apparatus, including: the device comprises a storage component and a processing component, wherein the storage component stores one or more computer instructions which are used for the processing component to call and execute;

the processing assembly is configured to:

acquiring a plurality of images to be processed; dividing an image to be processed into a plurality of area images aiming at any image to be processed; respectively extracting color histogram features of the plurality of region images; weighting the color histogram features of the plurality of region images to obtain image features of the image to be processed; classifying the images to be processed based on the image characteristics of the images to be processed to obtain a plurality of image sets; determining the image type of at least one image to be processed, which corresponds to each of the plurality of image sets; and determining the number of images of at least one image to be processed corresponding to each image category.

In the embodiment of the application, after extracting a plurality of clothing images, the clothing images can be divided into a plurality of area images for any one clothing image, and then, the color histogram feature of each area image can be extracted respectively, and the color histogram feature of each area image can embody the local color feature of the clothing image in the area; the color histograms of the plurality of area images are weighted to generate image features of the clothing image, and the obtained image features can realize the attention to the overall color features of the product image; in addition, since the texture of an image is actually generated by a change in pixel value, the histogram feature of each region image is actually obtained by counting the color histogram of each pixel point, and thus the image feature obtained by the color histogram weighting process of a plurality of region images can also characterize the image in terms of texture. And based on the image characteristics of the plurality of clothing images, an accurate classification result can be obtained when the plurality of clothing images are clustered. Because the image features of each clothing image can represent the color and texture features of the clothing image locally and integrally, the classification errors of the obtained multiple image sets are small, and the number of images of the multiple image sets corresponding to at least one clothing image can be accurately determined, so that the accurate monitoring of the clothing production process can be realized through the number of the images of the clothing images.

These and other aspects of the present application will be more readily apparent from the following description of the embodiments.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief description will be given below of the drawings that are needed in the embodiments or the prior art descriptions, and it is obvious that the drawings in the following description are some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 illustrates a flow chart of one embodiment of a garment image classification method provided herein;

FIG. 2 is a flow chart illustrating yet another embodiment of a garment image classification method provided herein;

FIG. 3 illustrates a flow chart of one embodiment of a picture classification method provided herein;

FIG. 4 is a schematic view of an embodiment of a clothing image classifying device according to the present application;

FIG. 5 is a schematic view of an embodiment of a garment image classifying apparatus according to the present application;

FIG. 6 is a schematic diagram illustrating the construction of one embodiment of a picture classification apparatus provided herein;

Fig. 7 is a schematic structural diagram of an embodiment of a picture classifying apparatus provided in the present application.

Detailed Description

In order to enable those skilled in the art to better understand the present application, the following description will make clear and complete descriptions of the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application.

In some of the flows described in the specification and claims of this application and in the foregoing figures, a number of operations are included that occur in a particular order, but it should be understood that the operations may be performed in other than the order in which they occur or in parallel, that the order of operations such as 101, 102, etc. is merely for distinguishing between the various operations, and that the order of execution is not by itself represented by any order of execution. In addition, the flows may include more or fewer operations, and the operations may be performed sequentially or in parallel. It should be noted that, the descriptions of "first" and "second" herein are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, and are not limited to the "first" and the "second" being different types.

The method and the device can be applied to digital factory management, and the production progress of the clothing can be known at any time by intelligently monitoring the production process of the clothing, so that the production efficiency is improved.

In the prior art, a camera can be used for image acquisition of clothing, image characteristics such as textures and outlines of the images are extracted, the acquired images are classified based on the image characteristics, and then the images with different types are obtained, so that the production progress of the clothing corresponding to the images with different types is determined. However, when an image is characterized by features such as texture and contour, the classification result of the image is not accurate enough, and classification errors are likely to occur. The inventor finds that when the image features such as the textures, the outlines and the like of the images are extracted, the obtained image features cannot well represent the local features because the whole image is directly extracted, and the extracted features are easily influenced by factors such as the shape, the color and the pattern of the clothing, are also easily influenced by factors such as illumination, exposure and the like, and further the extracted features are inaccurate, so that classification errors are caused.

Accordingly, the inventor proposes a technical solution of the present application, in the embodiment of the present invention, after extracting a plurality of clothing images, the clothing image may be divided into a plurality of area images for any one clothing image, and then, the color histogram feature of each area image may be extracted separately, and the color histogram features of the plurality of area images may be weighted, so as to obtain the image feature of the clothing image. Dividing the clothing image into a plurality of area images, extracting the color histogram of each area image to realize the extraction of local features of the clothing image, and carrying out weighting treatment on the color histogram features of the plurality of area images to realize the attention to the integral features of the clothing image, wherein the obtained image features of the clothing image can be comprehensively represented from a space angle; in addition, the histogram features can characterize the image in terms of color and texture and are not susceptible to the garment itself and the image acquisition environment. And further, the image characteristics of the clothing images are generated by using the color histogram weighting processing of the plurality of area images, and when the plurality of clothing images are clustered, accurate classification results can be obtained. The obtained multiple image sets have smaller classification errors, the image quantity of at least one clothing image corresponding to the multiple image sets can be accurately determined, and the accurate monitoring of the clothing production process can be realized through the image quantity of the clothing images.

The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

As shown in fig. 1, a flowchart of an embodiment of a clothing image classifying method according to an embodiment of the present invention may include the following steps:

101: a plurality of garment images is acquired.

Wherein, each clothing image can correspond to one clothing, and the clothing corresponding to each image is different.

Alternatively, in order to obtain accurate garment production quantities for production scheduling of different types of garments, the acquisition of garment images may be performed for garments that are produced and in a quality inspection procedure.

In practical application, since one factory may include a plurality of production lines, images acquired by different cameras may have different sizes, and for the same classification processing of images, a plurality of images acquired by different cameras may be normalized to obtain a plurality of clothing images. When normalizing the plurality of images, the aspect ratio of the original image can be maintained, and a plurality of clothing images with unchanged aspect ratio and normalized size can be obtained.

102: for any one of the garment images, the garment image is divided into a plurality of area images.

Any one of the plurality of clothing images may be selected, and the selected clothing image may be divided into a plurality of area images.

In some embodiments, to obtain the balanced division result, for any one of the clothing images, the clothing image may be divided into a plurality of area images according to the preset area position and the preset area size.

The region position may refer to a center position of the region image, and the region size may refer to an image size of the region image, specifically, may refer to a number of horizontal pixels and a number of vertical pixels of the region image.

In one possible implementation, in order to enable the clothing images to be uniformly divided, color features of different positions are acquired, and any clothing image may be divided into a plurality of area images according to the number of n×m. N may be equal to M, and the number of the lateral and longitudinal divisions of the clothing image is the same, and thus, the area size of each area image is the same. Wherein N is more than 1 and less than the number of the pixels of the clothing image. In practical application, N may take a value of 3, that is, the garment image may be uniformly divided into 9 area images with the same size.

103: and respectively extracting color histogram features of the plurality of region images.

The clothing image is segmented to obtain the color histogram feature of each area image respectively, so that the local feature of the clothing image is extracted, and the local feature of the clothing image is focused.

In order to improve the statistical efficiency, in practical application, histogram statistics can be directly performed on a plurality of pixel points of the area image, so as to obtain the color histogram feature of the area image. And respectively extracting color histogram features of the plurality of region images to obtain color histogram features respectively corresponding to the plurality of region images. 104: and weighting the color histogram features of the plurality of area images to obtain the image features of the clothing image.

Wherein the color histogram feature of each region image may be represented in the form of a color feature matrix.

As a possible implementation manner, the color histogram features of the plurality of area images may be weighted according to a predetermined weighting rule, so as to obtain the image features of the clothing image. The predetermined weighting rule may actually refer to that the color histogram feature of each area image is weighted according to the feature weight and the weighting position corresponding to each area image, so as to obtain the image feature of the clothing image.

In some embodiments, in order to equalize the features of each region to the image features of the garment image so that the obtained image features are more accurate, the weight of the color histogram features of each region image may be set to 1. When the color histogram feature of each area image is a color feature matrix, the weighting the color histogram features of the plurality of area images to obtain the image feature of the clothing image may include: determining a weighted sequence of color histogram features of each region image according to the position of each region image in the garment image; and transversely or longitudinally splicing the color feature matrixes according to the corresponding weighting sequences, wherein the obtained feature matrixes are the image features of the clothing image.

In practical application, in order to ensure that the local influence of the color histogram feature of each area image on the image feature of the clothing image is the same, an accurate image feature is generated, the feature weight of each area image can be set to be the same, for example, 1, and at this time, feature stitching can be directly performed on the color histogram features corresponding to a plurality of area images respectively, so as to obtain the image feature of the clothing image. For example, assuming that color histogram features corresponding to 3 adjacent area images A1, A2, A3 are B1, B2, B3, respectively, the features of the clothing images corresponding to the area images A1, A2, A3 may be B1B2B3.

105: and clustering the plurality of clothing images based on the image features of the plurality of clothing images to obtain a plurality of image sets.

Each of the garment images may represent a garment, and the image features of each of the garment images may represent the features of the garment, if the garment images are identical, indicating that the garments to which the garment images correspond are identical. For the same type of garment, the image features of its garment image are identical. Based on the image features of each of the plurality of clothing images, the plurality of clothing images are clustered to obtain a plurality of image sets, specifically, the image features of each of the plurality of clothing images may be obtained by dividing the clothing images with higher feature similarity into the same type of images, and obtaining a plurality of image sets.

Based on the image characteristics of each of the plurality of clothing images, clustering the plurality of clothing images by using a clustering algorithm to obtain a plurality of image sets.

In some embodiments, a clustering algorithm may be used to perform feature clustering on image features of each of the plurality of clothing images, and then, using clothing images corresponding to feature classes with the same image features as one image set, to obtain a plurality of image sets.

At least one clothing image can be corresponding to each image set, and the clothing corresponding to the at least one clothing image is the same type of clothing.

106: the number of images of the plurality of image sets corresponding to at least one garment image, respectively, is determined.

At least one clothing image can be corresponding to each image set, and the number of the at least one clothing image can be counted to obtain the number of images, namely, the number of images corresponding to each image set is obtained.

In some embodiments, after determining the number of images of the plurality of image sets corresponding to at least one garment image respectively, the plurality of image sets and the number of images corresponding to each image set may be further output, so as to facilitate the user to view.

In the embodiment of the invention, after a plurality of clothing images are extracted, the clothing images can be divided into a plurality of area images aiming at any one clothing image, and then, the color histogram characteristic of each area image can be extracted respectively, and the color histogram characteristic of each area image can embody the local color characteristic of the clothing image in the area; the color histograms of the plurality of area images are weighted to generate image features of the clothing image, and the obtained image features can realize attention to the overall color features of the product image.

In addition, since the texture of an image is actually generated by a change in pixel value, the histogram feature of each region image is actually obtained by counting the color histogram of each pixel point, and thus the image feature obtained by the color histogram weighting process of a plurality of region images can also characterize the image in terms of texture. And based on the image characteristics of the plurality of clothing images, an accurate classification result can be obtained when the plurality of clothing images are clustered. Because the image features of each clothing image can represent the color and texture features of the clothing image locally and integrally, the classification errors of the obtained multiple image sets are small, and the number of images of the multiple image sets corresponding to at least one clothing image can be accurately determined, so that the accurate monitoring of the clothing production process can be realized through the number of the images of the clothing images.

In practical applications, the color histogram features may comprise a color feature matrix, i.e. the color histogram features express the characteristics of the image in the form of a color feature matrix.

The weighting the color histogram features of the plurality of area images to obtain the image features of the clothing image may include:

and performing matrix splicing processing on the color feature matrixes of the region images in the transverse direction or the longitudinal direction to obtain the image features of the clothing image.

The transverse splicing treatment of any two color feature matrices may specifically refer to splicing treatment of any one row of the first color feature matrix with a row corresponding to the second color feature matrix, so as to obtain transverse splicing matrices of the two color feature matrices. For example, assume that the first color feature matrix is [101;100], the second color feature matrix is [202;200, performing transverse splicing on the two color feature matrixes to obtain a transverse splicing matrix [101202;100200].

The longitudinal stitching process of any two color feature matrices may specifically refer to performing longitudinal stitching on the first color feature matrix and the second color feature matrix based on the last row of the first color feature matrix and the first row of the second color feature matrix, so as to obtain longitudinal stitching matrices of the two color feature matrices. For example, assume that the first color feature matrix is [101;100], the second color feature matrix is [202;200, performing longitudinal splicing on the two color feature matrixes to obtain a longitudinal splicing matrix which can be [101;100;202;200].

As shown in fig. 2, a flowchart of still another embodiment of a clothing image classification method according to an embodiment of the present invention, which is different from the embodiment shown in fig. 1, is that after determining, in step 106, the image numbers of the plurality of image sets corresponding to at least one clothing image respectively, the method may further include:

201: and determining the type of clothing in which at least one clothing image corresponding to each of the plurality of image sets is located.

Each image set may correspond to at least one clothing image, and the clothing images belonging to the same image belong to the same clothing type, so that the at least one clothing image corresponding to each image set belongs to the same clothing type, and the clothing type in which the at least one clothing image corresponding to each of the plurality of image sets is located may be determined.

As an embodiment, the determining the garment type of the at least one garment image corresponding to each of the plurality of image sets may include:

determining at least one training image and an image feature of each training image; wherein each training image corresponds to a garment type;

determining training images matched with each image set based on image features of at least one garment image corresponding to the image sets respectively;

And the garment types of the training images corresponding to the image sets are taken as the garment types of at least one garment image corresponding to the image sets.

In order to make the feature expression modes of the images identical to realize feature matching of different images, the image features of the training image are obtained in the same manner as those of the embodiment shown in fig. 1 to 2.

Thus, the image features of each training image may be obtained by:

for any training image, dividing the training image into a plurality of training area images;

respectively extracting color histogram features of the plurality of training area images;

and weighting the color histogram features of the training area images to obtain the image features of the training images.

The region positions of each region image are different, and in order to make the representation modes of the image features of different clothing images identical, the color histogram features of the region images can be weighted according to the region positions of the regions to obtain the image features of the clothing images. As yet another embodiment, step 104 is: weighting the color histogram features of the plurality of area images to obtain image features of the garment image may include:

And weighting the color histogram features of the plurality of area images according to the area positions of the plurality of area images to obtain the image features of the clothing image.

The weighting processing according to the region positions of the plurality of region images may specifically refer to determining the weighting position of the region image according to the region position of any region image, and performing feature stitching on the region image according to the corresponding weighting position and the corresponding color histogram feature, so as to obtain the image feature of the clothing image when the feature stitching of the color histogram features of the plurality of region images is completed. When the color histogram feature is a color feature matrix, performing feature stitching on the color histogram feature may refer to performing matrix stitching on the color histogram matrix.

Since the overall color characteristics of different pixels can be represented by simply counting the pixel values of each pixel, but the characteristics of different pixels in different color spaces cannot be displayed, as a further embodiment, the extracting the color histogram characteristics of the plurality of area images in step 103 respectively may include:

and respectively extracting color histogram features of the plurality of region images in a plurality of color channels.

The color histogram features of each region image in a plurality of color channels may be extracted in turn. Color histogram features of the plurality of region images can be extracted based on the plurality of color channels, respectively.

Each color channel can represent different color characteristics, and the characteristics of different pixels in different color spaces can be obtained by converting the characteristics into corresponding color channels.

In order to obtain accurate color histogram features, as a further embodiment, the extracting color histogram features of the plurality of region images in a plurality of color channels respectively may include:

extracting channel histogram features of the region images corresponding to a plurality of color channels respectively for any one of the region images;

and weighting the channel histogram features corresponding to the color channels respectively to obtain the color histogram features of the region image.

Channel histogram features may refer to a matrix of channel features. Weighting the channel histogram features corresponding to the plurality of color channels respectively to obtain the color histogram features of the area image may include: and performing matrix splicing processing on the channel feature matrixes transversely or longitudinally to obtain color histogram features of the regional image. The weighting sequence of each color channel can be determined through the channel sequence of the color channels, and the feature matrix obtained by transversely or longitudinally splicing the feature matrices of the channels according to the weighting sequence is the color histogram feature of the regional image.

And extracting channel histogram features of the region image respectively corresponding to the plurality of color histogram channels, and weighting the plurality of channel histogram features to obtain the color histogram features of the region image. In practical application, for the weighting processing of the channel histogram features corresponding to the plurality of color channels, each channel histogram feature may be subjected to weighting processing to generate a color histogram of the area image.

In some embodiments, for any one of the plurality of area images, extracting channel histogram features of the area image corresponding to a plurality of color channels respectively includes:

for any one of the plurality of region images, respectively corresponding channel values of each pixel point of the region image in the plurality of color channels;

dividing each color channel into a plurality of histogram statistical intervals;

for any color channel, counting channel values of a plurality of pixel points of the regional image in the color channel, and respectively corresponding distribution numbers of a plurality of histogram statistical intervals corresponding to the color channel;

and determining channel histogram characteristics of the regional image in the color channel based on the distribution quantity of each histogram statistical interval.

In some embodiments, the clothing image classifying method may further include: and determining a plurality of color channels formed by the three color channels corresponding to the at least one color space model respectively.

The at least one color space module can be set in advance, a plurality of color channels can be obtained through three color channels corresponding to the at least one color space model respectively, and the plurality of color channels can be determined to be a plurality of color channels formed by the three color channels corresponding to the at least one color space model which is set in advance, so that the processing efficiency is improved.

For any one of the region images, channel histograms corresponding to the region images in a plurality of color channels can be extracted, specifically, pixel values of each of a plurality of pixels of the region images can be respectively converted into a plurality of color channels to obtain channel values corresponding to the plurality of color channels of each pixel, and for each color channel, channel values corresponding to the plurality of pixels of each region image in the channel can be obtained, and then channel histogram statistics can be performed for any one of the color channels.

Because of the large number of channels, if the data corresponding to each channel is counted one by one, an incremental calculation process is required, and the data dimension of the obtained channel histogram features is very high. Therefore, each color channel may be subjected to histogram statistical interval division so as to perform statistics on different color channels according to the histogram statistical interval.

In some possible designs, the at least one color space includes an RGB color space, an HSV color space, and a YcbCr color space;

wherein, the RGB color space corresponds to an R color channel, a G color channel and a B color channel, the HSV color space corresponds to an H color channel, an S color channel and a V color channel, and the YcbCr color space corresponds to a Y color channel, a Cb color channel and a Cv color channel.

For different color channels, each color channel may be divided into a plurality of histogram statistical bins according to the division times for its channel characteristics.

As one possible implementation, the R color channel may be divided into 24 histogram bins, the G color channel into 24 histogram bins, and the B color channel into 24 histogram bins, the HSV color space into 32 histogram bins, the S color channel into 32 histogram bins, and the V color channel into 8 histogram bins, the YcbCr color space into 24 histogram bins, the Cb color channel into 16 histogram bins, and the Cv color channel into 16 bins.

And the occurrence times of the channel values on the plurality of color channels corresponding to the plurality of pixel points in each statistical interval can be counted, so that the histogram feature value corresponding to each statistical interval can be obtained.

Taking the above-mentioned G color channel division as an example, after the G color channel is divided into 24 histogram statistical intervals, the number of occurrences of channel values corresponding to the G color channel in each statistical interval of a plurality of pixel points can be counted, so as to obtain the distribution numbers corresponding to the G color channel in the 24 statistical intervals, and obtain the channel histogram characteristic of the G color channel, where the characteristic is 1×24-dimensional data. Histogram features corresponding to R, B, H, S, V, Y, cb, cr color channels can be obtained, and the dimensions of the features are 1×24, 1×32, 1*8, 1×24, 1×16, and 1×16, respectively.

Then, the channel histogram features corresponding to the color channels may be weighted to obtain the color histogram features of the area image, and feature weighting is performed in this manner, for example, the channel histogram features corresponding to each channel are directly weighted to obtain a color feature histogram of 1×200 dimensions. If each clothing channel is divided into 9 area images on average, feature weighting processing is performed on color histogram features corresponding to the area images respectively, for example, simple feature stitching is performed, and image features with 1 x 1800 dimensions of clothing images can be obtained.

After the image features are used for clustering the plurality of clothing images based on the image features, clustering results of the plurality of clothing images can be output, so that a user can check the clothing images conveniently. As yet another embodiment, after determining the number of images of the plurality of image sets corresponding to at least one clothing image, the method further includes:

sorting the number of images to determine an output order of each image set;

and sequentially outputting at least one clothing image corresponding to the image sets according to the respective output sequence.

Alternatively, sorting the number of images to determine the output order of each image set may specifically refer to sorting the number of images up or down by number size to determine the output order of each image set.

As shown in fig. 3, a flowchart of still another embodiment of an image classification method according to an embodiment of the present invention may include the following steps:

301: a plurality of images to be processed are acquired.

302: for any one image to be processed, the image to be processed is divided into a plurality of area images.

303: and respectively extracting color histogram features of the plurality of region images.

304: and weighting the color histogram features of the plurality of region images to obtain the image features of the image to be processed.

305: classifying the images to be processed based on the image characteristics of the images to be processed to obtain a plurality of image sets.

306: and determining the image type of at least one image to be processed, which corresponds to the plurality of image sets respectively.

307: and determining the number of images of at least one image to be processed corresponding to each image category.

The image classification method provided by the embodiment of the invention is the same as the image processing mode used in the clothing image classification method shown in fig. 1-2, and is not repeated here.

In the embodiment of the invention, after a plurality of images to be processed are extracted, the images to be processed can be divided into a plurality of area images aiming at any one of the images to be processed, then the color histogram features of each area image can be extracted respectively, and the color histogram features of the plurality of area images are weighted to obtain the image features of the images to be processed. The color histogram features of the region images can embody the local features of the images to be processed, and the weighting treatment of the color histogram features of a plurality of region images can realize the attention of the integral features of the images to be processed; in addition, the histogram features can characterize the image in terms of color and texture and are not susceptible to the garment itself and the image acquisition environment. And then, the color histogram weighting processing of the plurality of area images is used for generating the image characteristics of the images to be processed, and when the plurality of images to be processed are clustered, an accurate classification result can be obtained. The obtained multiple image sets have smaller classification errors, the image quantity of at least one image to be processed corresponding to the multiple image sets can be accurately determined, and the accurate monitoring of the clothing production process can be realized through the image quantity of the image to be processed.

As shown in fig. 4, a schematic structural diagram of still another embodiment of a clothing image classifying device according to an embodiment of the present invention may include the following modules:

a first acquiring module 401, configured to acquire a plurality of clothing images;

a first dividing module 402, configured to divide, for any one clothing image, the clothing image into a plurality of area images;

a first extraction module 403, configured to extract color histogram features of the plurality of area images respectively;

a first weighting module 404, configured to perform weighting processing on the color histogram features of the plurality of area images, so as to obtain image features of the clothing image;

a first clustering module 405, configured to cluster the plurality of clothing images based on image features of the plurality of clothing images, to obtain a plurality of image sets;

a first determining module 406 is configured to determine a number of images of the plurality of image sets corresponding to at least one clothing image, respectively.

In the embodiment of the application, the color histogram features of the area images can embody the local features of the clothing images, and the color histogram features of the area images are weighted so as to pay attention to the integral features of the clothing images; in addition, the histogram features can characterize the image in terms of color and texture and are not susceptible to the garment itself and the image acquisition environment. And further, the image characteristics of the clothing images are generated by using the color histogram weighting processing of the plurality of area images, and when the plurality of clothing images are clustered, accurate classification results can be obtained. The obtained multiple image sets have smaller classification errors, the image quantity of at least one clothing image corresponding to the multiple image sets can be accurately determined, and the accurate monitoring of the clothing production process can be realized through the image quantity of the clothing images.

As an embodiment, the apparatus shown in fig. 4 may further include:

and the fourth determining module is used for determining the type of clothing where at least one clothing image corresponding to the plurality of image sets respectively is located.

As yet another embodiment, the apparatus shown in fig. 4 may further include:

the sorting module is used for sorting the image quantity to determine the output sequence of each image set; the output module is used for sequentially outputting at least one clothing image corresponding to the image sets according to the respective output sequence.

In order to make the representation modes of the image features of different clothing images identical, the color histogram features of the area images may be weighted according to the area positions to obtain the image features of the clothing images, and as a further embodiment, the first weighting module includes:

And the first weighting unit is used for weighting the color histogram features of the plurality of area images according to the area positions of the plurality of area images to obtain the image features of the clothing image.

Since the color overall characteristics of different pixels can be represented by simply counting the pixel value of each pixel, but the characteristics of different pixels in different color spaces cannot be displayed, as a further embodiment, the first extraction module includes:

and the first extraction unit is used for respectively extracting the color histogram features of the plurality of region images in a plurality of color channels.

In order to obtain accurate color histogram features, as a further embodiment, the first extraction unit comprises:

a first extraction subunit, configured to extract channel histogram features of any one of the plurality of region images, where the channel histogram features correspond to a plurality of color channels of the region image respectively;

and the first weighting subunit is used for carrying out weighting processing on channel histogram features corresponding to the plurality of color channels respectively to obtain the color histogram features of the regional image.

In certain embodiments, the first extraction subunit may be specifically configured to include:

In certain embodiments, further comprising:

the channel determining module is used for determining a plurality of color channels formed by three color channels corresponding to at least one color space model respectively.

As one embodiment, the color histogram feature comprises a color feature matrix;

the first weighting module may include:

and the second weighting module is used for performing matrix splicing processing on the color feature matrixes of the region images in the transverse direction or the longitudinal direction to obtain the image features of the clothing image.

The apparatus shown in fig. 4 may perform the clothing image classification method described in the embodiment shown in fig. 1 to 2, and its implementation principle and technical effects will not be repeated. The specific manner in which the respective modules and units of the apparatus for classifying the clothing images in the above embodiments perform operations has been described in detail in the embodiments related to the method, and will not be described in detail herein.

In addition, the embodiment of the application further provides a computer readable storage medium, and a computer program is stored, and when the computer program is executed by a computer, the clothing image classification method of the embodiment shown in fig. 1-2 can be realized.

As shown in fig. 5, a schematic structural diagram of still another embodiment of a garment image classifying apparatus according to an embodiment of the present invention may include: a storage component 501 and a processing component 502, said storage component 501 storing one or more computer instructions for said processing component 502 to call and execute;

The processing component 502 may be configured to:

As an example, the processing component 501 may also be configured to:

and determining the type of clothing in which at least one clothing image corresponding to each of the plurality of image sets is located.

As yet another example, the processing assembly 501 may also be configured to:

sorting the number of images to determine an output order of each image set;

Since the overall color characteristics of different pixels can be represented by simply counting the pixel value of each pixel, but the characteristics of different pixels in different color spaces cannot be displayed, as a further embodiment, the processing component performs weighting processing on the color histogram characteristics of the multiple area images, and the image characteristics of the obtained clothing image may specifically be:

In order to obtain accurate color histogram features, as a further embodiment, the processing component extracts color histogram features of the plurality of area images respectively, specifically may be:

In some embodiments, the processing component extracts color histogram features of the plurality of area images in a plurality of color channels, where the color histogram features may be: extracting channel histogram features of the region images corresponding to a plurality of color channels respectively for any one of the region images;

In some embodiments, the processing component may specifically extract, for any one of the plurality of area images, channel histogram features of the area image corresponding to the plurality of color channels respectively, where the channel histogram features may be:

In some embodiments, the processing component 501 may also be configured to:

and determining a plurality of color channels formed by the three color channels corresponding to the at least one color space model respectively.

In some possible designs, the at least one color space determined by the processing component includes an RGB color space, an HSV color space, and a YcbCr color space;

The apparatus shown in fig. 5 may perform the clothing image classification method described in the embodiment shown in fig. 1 to 2, and its implementation principle and technical effects will not be repeated. The relevant specific manner in which the processing components of the apparatus for classifying a garment image in the above-described embodiments has been described in detail in relation to the embodiments of the method, and will not be described in detail here.

As shown in fig. 6, a schematic structural diagram of still another embodiment of an image classification apparatus according to an embodiment of the present invention may include the following modules:

A second acquiring module 601, configured to acquire a plurality of images to be processed;

a second dividing module 602, configured to divide, for any image to be processed, the image to be processed into a plurality of area images;

a second extraction module 603, configured to extract color histogram features of the plurality of area images respectively;

a second weighting module 604, configured to perform weighting processing on the color histogram features of the plurality of area images, so as to obtain image features of the image to be processed;

a second clustering module 605, configured to classify the plurality of images to be processed based on image features of the plurality of images to be processed, to obtain a plurality of image sets;

a second determining module 606, configured to determine an image type to which at least one image to be processed corresponding to each of the plurality of image sets belongs;

a third determining module 607 is configured to determine the number of images of the at least one image to be processed corresponding to each image category.

In the embodiment of the invention, the color histogram features of the region images can embody the local features of the images to be processed, and the weighting treatment of the color histogram features of a plurality of region images can realize the attention of the integral features of the images to be processed; in addition, the histogram features can characterize the image in terms of color and texture and are not susceptible to the garment itself and the image acquisition environment. And then, the color histogram weighting processing of the plurality of area images is used for generating the image characteristics of the images to be processed, and when the plurality of images to be processed are clustered, an accurate classification result can be obtained. The obtained multiple image sets have smaller classification errors, the image quantity of at least one image to be processed corresponding to the multiple image sets can be accurately determined, and the accurate monitoring of the clothing production process can be realized through the image quantity of the image to be processed.

The apparatus shown in fig. 6 may perform the image classification method described in the embodiment shown in fig. 3, and its implementation principle and technical effects are not repeated. The specific manner in which the respective modules and units of the image classification apparatus perform operations in the above embodiments has been described in detail in the embodiments related to the method, and will not be described in detail herein.

In addition, the embodiment of the application further provides a computer readable storage medium, and a computer program is stored, and when the computer program is executed by a computer, the image classification method of the embodiment shown in fig. 3 can be realized.

As shown in fig. 7, a schematic structural diagram of still another embodiment of an image classification apparatus according to an embodiment of the present invention may include: a storage component 701 and a processing component 702, the storage component 701 storing one or more computer instructions for the processing component 702 to call and execute;

the processing component 702 is configured to:

The apparatus shown in fig. 7 may perform the image classification method described in the embodiment shown in fig. 3, and its implementation principle and technical effects will not be described again. The relevant specific manner of the processing components in the image classification apparatus in the above embodiments has been described in detail in the embodiments related to the method, and will not be described in detail here.

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

The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting thereof; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.

Claims

1. A method of garment image classification, comprising:

acquiring a plurality of clothing images;

the channel histogram features corresponding to the color channels are spliced, so that the color histogram features of the regional image are obtained;

according to the region positions of the region images, performing feature stitching on the color histogram features of the region images to obtain the image features of the clothing images;

the number of images of the plurality of image sets corresponding to at least one garment image, respectively, is determined.

2. The method as recited in claim 1, further comprising:

3. The method of claim 1, wherein after determining the number of images of the plurality of image sets corresponding to at least one garment image, respectively, further comprises:

sorting the number of images to determine an output order of each image set;

4. The method of claim 1, wherein extracting channel histogram features of the region image corresponding to a plurality of color channels, respectively, for any one of the plurality of region images comprises:

5. The method as recited in claim 4, further comprising:

6. The method of claim 5, wherein the at least one color space comprises an RGB color space, an HSV color space, and a YcbCr color space;

7. The method of claim 1, wherein the color histogram feature comprises a color feature matrix;

The feature stitching of the color histogram features of the plurality of region images includes:

and performing transverse or longitudinal matrix splicing on the color feature matrix of the regional image.

8. An image classification method, comprising:

acquiring a plurality of images to be processed;

according to the region positions of the region images, performing feature stitching on the color histogram features of the region images to obtain image features of the image to be processed;

9. A garment image classification device, comprising:

the first extraction module is used for extracting channel histogram features of the region images corresponding to a plurality of color channels respectively aiming at any one of the region images; the channel histogram features corresponding to the color channels are spliced, so that the color histogram features of the regional image are obtained;

the first weighting module is used for carrying out characteristic stitching on the color histogram characteristics of the plurality of area images according to the area positions of the area images to obtain the image characteristics of the clothing image;

10. An image classification apparatus, comprising:

the second extraction module is used for extracting channel histogram features of the region images corresponding to a plurality of color channels respectively aiming at any one of the region images; the channel histogram features corresponding to the color channels are spliced, so that the color histogram features of the regional image are obtained;

the second weighting module is used for carrying out characteristic stitching on the color histogram characteristics of the plurality of regional images according to the regional positions of the regional images to obtain the image characteristics of the image to be processed;

11. A garment image classification device, comprising: the device comprises a storage component and a processing component, wherein the storage component stores one or more computer instructions which are used for the processing component to call and execute;

the processing assembly is configured to:

acquiring a plurality of clothing images; dividing a garment image into a plurality of area images for any one garment image; extracting channel histogram features of the region images corresponding to a plurality of color channels respectively for any one of the region images; the channel histogram features corresponding to the color channels are spliced, so that the color histogram features of the regional image are obtained; according to the region positions of the region images, performing feature stitching on the color histogram features of the region images to obtain the image features of the clothing images; clustering the plurality of clothing images based on the image features of the plurality of clothing images to obtain a plurality of image sets; the number of images of the plurality of image sets corresponding to at least one garment image, respectively, is determined.

12. An image classification apparatus, characterized by comprising: the device comprises a storage component and a processing component, wherein the storage component stores one or more computer instructions which are used for the processing component to call and execute;

The processing assembly is configured to:

acquiring a plurality of images to be processed; dividing an image to be processed into a plurality of area images aiming at any image to be processed; extracting channel histogram features of the region images corresponding to a plurality of color channels respectively for any one of the region images; the channel histogram features corresponding to the color channels are spliced, so that the color histogram features of the regional image are obtained; according to the region positions of the region images, performing feature stitching on the color histogram features of the region images to obtain image features of the image to be processed; classifying the images to be processed based on the image characteristics of the images to be processed to obtain a plurality of image sets; determining the image type of at least one image to be processed, which corresponds to each of the plurality of image sets; and determining the number of images of at least one image to be processed corresponding to each image category.