CN113066039A

CN113066039A - Method and device for recognizing image subject

Info

Publication number: CN113066039A
Application number: CN201911288847.3A
Authority: CN
Inventors: 赵墨农
Original assignee: Beijing Wodong Tianjun Information Technology Co Ltd
Current assignee: Beijing Wodong Tianjun Information Technology Co Ltd
Priority date: 2019-12-12
Filing date: 2019-12-12
Publication date: 2021-07-02

Abstract

The invention discloses a method and a device for recognizing an image subject, and relates to the technical field of computers. One embodiment of the method comprises: performing region growing on the picture to be processed, and acquiring growing regions containing non-backgrounds; setting a target area in the picture to be processed, and screening target growing areas with intersection with the target area from each growing area; the target area is used for indicating the range of the area where the main body is located in the picture to be processed; and determining a main body image in the picture to be processed according to each target growing region. The method and the device can automatically complete image subject recognition, and avoid the problems of manpower consumption, low speed, insufficient response speed to new images, incapability of reaching pixel levels and the like caused by manual processing.

Description

Method and device for recognizing image subject

Technical Field

The invention relates to the technical field of computers, in particular to a method and a device for recognizing an image subject.

Background

The subject recognition of images is a pre-algorithm of a plurality of image algorithms, and the current stage subject recognition method mainly comprises the following steps: machine learning, an automatic labeling system, and the separation of a main body from a background by adopting modes of background subtraction, binarization and the like. Machine learning requires annotation data, which is mostly done manually. Automatic labeling systems are usually semi-manual, such as manually selecting contour points on the contour of an object in a picture, and generating a labeled picture from the contour points by a program. The main body and background separation is realized by adopting the modes of background subtraction, binarization and the like, and the method is mainly suitable for any picture.

In the process of implementing the invention, the inventor finds that at least the following problems exist in the prior art:

the manual processing consumes manpower, has low speed, has insufficient reaction speed on new images and often cannot reach the pixel level;

the method for realizing the main body and background separation by adopting the modes of background subtraction, binarization and the like is suitable for any picture, and has the following defects: the calculated amount is large; the edge processing is not accurate enough; the main body of the picture which is really outstanding cannot be accurately found out.

Disclosure of Invention

In view of this, embodiments of the present invention provide a method and an apparatus for image subject recognition, which can automatically complete image subject recognition, and avoid the problems of manpower consumption, slow speed, insufficient response speed to a new image, often failing to reach a pixel level, and the like caused by manual processing.

To achieve the above object, according to an aspect of an embodiment of the present invention, there is provided an image subject recognition method including:

performing region growing on the picture to be processed, and acquiring growing regions containing non-backgrounds;

setting a target area in the picture to be processed, and screening target growing areas with intersection with the target area from each growing area; the target area is used for indicating the range of the area where the main body is located in the picture to be processed;

and determining a main body image in the picture to be processed according to each target growing region.

Optionally, before performing region growing on the picture to be processed, the method further includes: screening the pictures to be processed from the pictures according to the following steps:

respectively making first rays in positive and negative directions of a preset angle from the central point of the picture, and judging whether a straight line where the first rays are located meets a preset condition; if so, taking the straight line of the first ray as a main image frame of the picture; otherwise, moving the two first rays along the directions of (a preset angle +90 degrees) and (a preset angle-90 degrees) respectively until the straight line where the first rays are located after moving meets the preset condition, and taking the straight line where each first ray is located after moving as a main image frame of the picture respectively; the preset condition is that the number of continuous background pixel points including the first ray end point is greater than or equal to a preset number threshold;

replacing one preset angle with another preset angle, then repeatedly executing the steps until a preset stop condition is reached, and taking the picture which can reach the preset stop condition in all the pictures as the picture to be processed;

the preset stop condition is as follows: the straight line where the ray is located is not blocked by the subject pixel points, and two intersections are generated between the straight line and other subject image frames.

Optionally, the method of the embodiment of the present invention further includes: determining the number of the frames of the preset main image, and recording as M; and each angle equally divided by M/2 at 180 degrees is used as each preset angle.

Optionally, performing region growing on the picture to be processed, and acquiring each growing region including the non-background, including:

searching a pixel point containing a non-background from the picture to be processed as a non-background pixel point, and putting the non-background pixel point into a first set; traversing adjacent pixel points of the non-background pixel points, and putting the adjacent pixel points containing the non-background into the first set; taking the adjacent pixel points put into the first set as new non-background pixel points to increase until all the non-background pixel points in the adjacent pixel points put into the first set are put into the first set to obtain an increasing area;

setting the pixel value of a pixel point corresponding to the growing region in the picture to be processed as a background pixel value, and then executing the step of obtaining the growing region to obtain another growing region; and repeating the step until all pixel points in the picture to be processed are set as background pixel values, and obtaining each growing region containing non-background.

Optionally, determining a subject image in the to-be-processed picture according to each target growing region includes:

setting pixel values of all pixel points in the picture to be processed as background pixel values, and then fusing the background pixel values with the target growing region to obtain a candidate picture;

searching a pixel point containing a background from the candidate picture as a background pixel point, and putting the background pixel point into a second set; traversing adjacent pixel points of the background pixel points, and putting the adjacent pixel points containing the background into a second set; taking the adjacent pixel points put into the second set as new background pixel points to increase until all the background-containing pixel points in the adjacent pixel points put into the background pixel points of the second set are put into the second set to obtain a background area;

and filtering pixel points corresponding to the background area in the candidate picture, wherein each filtered target growing area is used as a main image.

Optionally, the method of the embodiment of the present invention further includes: for any pixel point, judging whether the any pixel point is a non-background pixel point or a background pixel point according to the following steps:

determining the sum of the RGB values of any pixel point, and then judging the relation between the sum of the RGB values of any pixel point and the lowest value and the highest value of the preset error; wherein the content of the first and second substances,

if the sum of RGB of any pixel point is less than or equal to the preset error lowest value, judging that any pixel point is a non-background pixel point;

if the sum of RGB of any pixel point is larger than the preset error maximum value, judging that any pixel point is a background pixel point;

if the sum of the RGB of any pixel point is larger than the preset error minimum value and smaller than or equal to the preset error maximum value, adjusting the preset error minimum value based on the sum of the RGB of eight adjacent pixel points of any pixel point to obtain a corrected error minimum value; and judging whether the sum of RGB of any pixel point is less than or equal to the minimum value of the correction error, if so, judging that the any pixel point is a non-background pixel point, and otherwise, judging that the any pixel point is a background pixel point.

Optionally, the preset error minimum is adjusted according to the following formula:

Z′＝Z+D×Y

in the formula, Z' represents the minimum value of the correction error, Z represents the minimum value of the preset error, D represents the sum of RGB of eight adjacent pixel points, and Y represents the increase of the preset error.

Optionally, after determining the subject image in the to-be-processed picture according to each target growing region, the method further includes:

and fusing the main body image and a preset background image to obtain a marked image sample corresponding to the picture to be processed.

According to a second aspect of the embodiments of the present invention, there is provided an apparatus for image subject recognition, including:

the region growing unit is used for performing region growing on the picture to be processed and acquiring each growing region containing the non-background;

the main body judging unit is used for setting a target area in the picture to be processed and screening target growing areas with intersection with the target area from each growing area; the target area is used for indicating the range of the area where the main body is located in the picture to be processed;

and the main body determining unit is used for determining a main body image in the picture to be processed according to each target growing region.

Optionally, the image processing apparatus according to the embodiment of the present invention further includes an image determining unit, configured to: before the region growth of the picture to be processed, screening the picture to be processed from each picture according to the following steps:

Optionally, the picture determination unit is further configured to: determining the number of the frames of the preset main image, and recording as M; and each angle equally divided by M/2 at 180 degrees is used as each preset angle.

Optionally, the region growing unit performs region growing on the picture to be processed, and acquires each growing region including a non-background, including:

Optionally, the determining, by the subject determining unit, a subject image in the to-be-processed picture according to each of the target growing regions includes:

Optionally, the region growing unit or the subject determination unit is further configured to:

for any pixel point, judging whether the any pixel point is a non-background pixel point or a background pixel point according to the following steps:

Z′＝Z+D×Y

Optionally, the apparatus for image subject identification according to the embodiment of the present invention further includes: a sample labeling unit for: and after determining a main image in the picture to be processed according to each target growing region, fusing the main image with a preset background image to obtain a labeled image sample corresponding to the picture to be processed.

According to a third aspect of embodiments of the present invention, there is provided an electronic device for image subject recognition, comprising:

one or more processors;

a storage device for storing one or more programs,

when the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the method provided by the first aspect of the embodiments of the present invention.

According to a fourth aspect of embodiments of the present invention, there is provided a computer readable medium, on which a computer program is stored, which when executed by a processor, implements the method provided by the first aspect of embodiments of the present invention.

One embodiment of the above invention has the following advantages or benefits: according to the method and the device, the growth area is obtained through area growth, the subject image is determined according to the target growth area with intersection with the target area, the image subject identification can be automatically completed, and the problems that manpower is consumed, the speed is low, the response speed to a new image is insufficient, the pixel level cannot be achieved and the like caused by manual processing are solved. By determining the target area in the picture to be processed, the main body of the picture which really needs to be highlighted can be accurately found out. The image to be processed is screened by the provided judging method, and the calculated amount of image main body identification in the subsequent steps can be reduced by adopting the provided region growing method and the region growing judging criterion, so that the accuracy of edge processing is improved. Through fusing the determined main body image with the preset background image, a labeled image sample can be obtained, a sample is provided for machine learning, and the problems that manpower is consumed, speed is low, the reaction speed to a new image is insufficient, the pixel level cannot be achieved frequently and the like caused by manually labeling the machine-learned sample are solved.

Further effects of the above-mentioned non-conventional alternatives will be described below in connection with the embodiments.

Drawings

The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:

fig. 1 is a schematic diagram of a main flow of a method of image subject recognition in an embodiment of the present invention;

FIG. 2 is a schematic diagram of a main flow of a method for screening a picture to be processed according to an embodiment of the present invention;

FIG. 3 is a diagram illustrating a main image frame according to an embodiment of the present invention;

FIG. 4 is a schematic diagram illustrating a main flow of a method for performing region growing on a picture to be processed according to an embodiment of the present invention;

FIG. 5 is a diagram illustrating a region growing criterion according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a main flow of reverse region growing on a target growing region in the embodiment of the present invention;

fig. 7 is a schematic diagram of the main components of an apparatus for image subject recognition according to an embodiment of the present invention;

FIG. 8 is an exemplary system architecture diagram in which embodiments of the present invention may be employed;

fig. 9 is a schematic structural diagram of a computer system suitable for implementing a terminal device or a server according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

According to an aspect of an embodiment of the present invention, there is provided a method of image subject recognition.

Fig. 1 is a schematic diagram of a main flow of a method for image subject identification according to an embodiment of the present invention, and as shown in fig. 1, the method for image subject identification includes:

s101, performing region growing on a picture to be processed, and acquiring growing regions containing non-backgrounds;

step S102, setting a target area in the picture to be processed, and screening target growing areas with intersection with the target area from each growing area; the target area is used for indicating the range of the area where the main body is located in the picture to be processed;

step S103, determining a main body image in the picture to be processed according to each target growing region.

The method is suitable for processing the picture with pure color background and the main body surrounded by the background. The method has the advantages that the growth area is obtained through area growth, the subject image is determined according to the target growth area with intersection with the target area, the image subject identification can be automatically completed, and the problems that manpower is consumed, the speed is low, the response speed to a new image is insufficient, the pixel level cannot be achieved and the like due to manual processing are solved.

The setting of the target area depends on the characteristics of the picture to be processed. Taking a commodity display picture in the e-commerce field as an example, in the commodity display picture, a main body to-be-displayed commodity is often placed at a position close to the middle of an image. Therefore, a target area can be set in the middle area of the picture to be processed. Of course, when the main body is usually disposed at other positions in other pictures to be processed, such as the upper left corner, the lower right corner, etc., a target region may be disposed in the corresponding region of the picture to be processed. The size of the target area can be selectively determined according to actual conditions, for example, the length and width of the target area are respectively set to be 5% of the length and width of the picture to be processed. In this step, all the growth regions obtained in step S101 are traversed, each growth region is determined, and a growth region having an intersection with the target region is screened as a target growth region. In general, one target growth area corresponds to one subject, and if the situation corresponding to two or more subjects occurs, the screening can be performed through multi-subject competition. The specific implementation manner of the multi-subject election can be selectively determined according to the actual situation, and details are not described here. The method and the device can accurately find out the main body of the picture which is really outstanding by determining the target area in the picture to be processed.

The method for identifying the image main body is particularly suitable for the picture to be processed with the background color being pure color. Optionally, before performing region growing on the picture to be processed, the method further includes: screening the pictures to be processed from the pictures according to the following steps:

Fig. 2 is a schematic diagram of a main flow of a method for screening a to-be-processed picture according to an embodiment of the present invention. For any picture, judging whether the picture can be used as a picture to be processed according to the following steps:

step S201, two first rays are made from the center point of the picture to the positive and negative directions of the ith preset angle; i represents a positive integer, and under the initial condition, i is 1; skipping to step S202;

step S202, judging whether the number of continuous background pixel points including a first ray end point in a straight line where a first ray is located is larger than or equal to a preset number threshold or not; if yes, jumping to step S204; otherwise, jumping to step S203;

step S203, moving one first ray along the direction of (the preset angle is plus 90 degrees), moving another first ray along the direction of (the preset angle is minus 90 degrees), and then jumping to the step S202;

step S204, taking a straight line where the first ray is located as a main image frame of the picture; jumping to step S205;

step S205, judging whether a preset stop condition is reached; if yes, go to step S206; otherwise, i +1, and then jumping to step S201;

and step S206, judging that the picture can be used as a picture to be processed, and ending the process.

The background pixel points refer to pixel points containing the background. The image main body recognition is carried out based on the picture to be processed with the pure color background, the calculated amount of the image main body recognition in the subsequent steps can be reduced, and the recognition speed is greatly improved. The method has the advantages that the image main body identification is carried out on the basis of the picture to be processed with the pure-color background, so that the marked image sample is obtained, the speed of marking the sample can be greatly improved, and the marked image sample can reach the pixel level.

The size and the number of the preset angles can be selectively determined according to actual conditions. For example, the preset angles are 4, 0 °, 90 °, 45 °, 135 °, respectively. Optionally, the image processing method according to the embodiment of the present invention further includes: determining the number of the frames of the preset main image, and recording as M; and each angle equally divided by M/2 at 180 degrees is used as each preset angle. Illustratively, the number X of the preset subject image borders is 8, and X/2 is 4, so the preset angle is 0 °, 90 °, 45 °, and 135 ° where 180 ° is equally divided by 4.

If the number of the borders of the preset main image is increased and the stop condition in the screening process shown in fig. 2 is removed, the boundary of the main image can be determined to be closer to the main image itself. But cannot fit perfectly to the subject image boundaries due to subject irregularities. If the number of the frames of the preset main body image is increased without removing the stop condition in the screening process shown in fig. 2, more to-be-processed images can be screened out, but the calculation amount for performing main body identification and labeling based on the to-be-processed images obtained by screening is increased.

Whether a picture can be used as the picture to be processed in the embodiment for subsequent main image identification and labeling is determined through screening, and in order to reduce the calculation amount, the number of the frames of the preset main image can be set to be a smaller value, for example, 8, so that a considerable number of pictures to be processed can be determined under the condition of small calculation amount.

Each preset angle is determined in an equal division mode, and the number of actual frames is convenient to adjust. It should be noted that, although the number of the frames of the preset main image is determined in this example, the number of the frames of the main image that is finally obtained is not represented, and is the preset number. When the screening is performed based on the method shown in fig. 2, if the predetermined stop condition is reached, the process is stopped even if the number of frames of the obtained main image is smaller than the predetermined number.

Fig. 3 is a schematic diagram of a main image frame according to an embodiment of the present invention, in which a1, a2, A3, a4, and a5 each represent a main image frame. In this example, two subject image frames are formed by radiating rays in positive and negative directions of 0 ° and shifting the subject image frames in directions of 0 ° +90 ° -90 ° and shifting the subject image frames in directions of 0 ° -90 °, which are denoted as a1 and a 2. Two subject image frames, denoted as A3 and a4, are formed by radiating rays in the plus and minus directions of 90 ° and shifting the subject image frames in the directions of 90 ° + 180 ° and 90 ° -0 °. At this time, a1 and A3 are blocked by the main body pixel points in the picture respectively, and cannot form two intersections with the borders of other main body images. Therefore, the next ray is made in the positive and negative directions of 45 °, and the ray is shifted in the direction of 45 ° +90 ° + 135 ° to form a frame of the main image, which is denoted as a 5. At this time, each of a1, a2, A3, a4 and a5 forms an intersection with the other two subject image borders, a stop condition is reached, the loop exits, and the subject in the picture is judged to be surrounded by the background and can be used as the picture to be processed.

Fig. 4 is a schematic diagram of a main flow of a method for performing region growing on a picture to be processed in an embodiment of the present invention, and as shown in fig. 4, the main flow for performing region growing is:

step S401, a pixel point containing a non-background is searched from the picture to be processed to serve as a non-background pixel point, and the non-background pixel point is placed into a first set; skipping to step S402;

s402, traversing adjacent pixel points of the non-background pixel points, and putting the adjacent pixel points containing the non-background into the first set; skipping to step S403;

step S403, taking the adjacent pixel points put into the first set as new non-background pixel points; skipping to step S404;

s404, judging whether all the pixels containing non-background in the adjacent pixels are put into a first set; if yes, jumping to step S405; otherwise, jumping to step S402;

step S405, judging whether all pixel points in the picture to be processed are set as background pixel values or not; if yes, jumping to step S406; otherwise, x +1, skipping to step S401;

and step S406, obtaining each growing region containing non-background.

The non-background pixels are pixels except for the background pixels. In the practical application process, when a pixel point is judged not to be a background pixel point, the pixel point can be judged to be a non-background pixel point. The background pixel points refer to pixel points containing the background. The background pixel value refers to the pixel value of the background pixel point. The regions formed in each first set are one growth region, and each growth region corresponds to one body. The region growing method provided by the invention can find out all growing regions containing non-background.

Optionally, as shown in fig. 6, for each target growth region, determining the subject image in the to-be-processed picture according to the target growth region includes:

s601, setting pixel values of all pixel points in the picture to be processed as background pixel values, and then fusing the background pixel values with the target growing region to obtain a candidate picture;

step S602, a pixel point containing a background is searched from the candidate picture as a background pixel point, and the background pixel point is put into a second set;

step S603, traversing adjacent pixel points of the background pixel points, and putting the adjacent pixel points containing the background into a second set;

step S604, taking the adjacent pixel point put into the second set as a new background pixel point;

step S605, judging whether all the pixels containing the background are put into the second set or not among the adjacent pixels of the background pixels put into the second set; if yes, obtaining a background area, and then jumping to the step S606; otherwise, jumping to step S603;

step S606, filtering the pixel points corresponding to the background regions in the candidate pictures, and taking each filtered candidate picture as a main image, thereby ending the process.

According to the method, the background pixel points in the candidate picture can be removed through the increase of the reverse region, and the accuracy of edge processing is further improved.

Optionally, for any pixel point, judging whether the any pixel point is a non-background pixel point or a background pixel point according to the following steps:

The values of the preset error minimum and the preset error maximum can be selectively determined according to the actual situation, and are not described herein again. According to the area growth judgment criterion, whether a certain pixel point is a main pixel point or a background pixel point can be accurately judged, and the accuracy of edge processing is improved.

Z′＝Z+D×Y

Exemplarily, as shown in fig. 5, the step of determining whether a pixel is a non-background pixel includes:

s501, determining the sum of the RGB values of the pixel point, and recording as C;

step S502, judging the relation between the sum C of the RGB values of the pixel point and a preset error minimum value (marked as Z) and a preset error maximum value (marked as X); if C is larger than X, jumping to step S503; if C is less than or equal to Z, jumping to step S504; if Z is more than C and less than or equal to X, jumping to step S505;

step S503, judging that the pixel point is a background pixel point;

step S504, judging that the pixel point is a non-background pixel point;

step S505, Z' ═ Z + D × Y, and then jump to step S506;

step S506, judging whether Z' is equal to or less than C; if yes, jumping to step S507; otherwise, jumping to step S508;

step S507, judging that the pixel point is a non-background pixel point;

step S508, determining that the pixel is a background pixel.

The preset error minimum value is adjusted based on the sum of RGB of the eight adjacent pixel points, so that the accuracy of main body identification and the accuracy of edge processing can be improved.

Optionally, after determining the subject image in the to-be-processed picture according to each target growing region, the method further includes: and fusing the main body image and a preset background image to obtain a marked image sample corresponding to the picture to be processed. When image fusion is performed, the subject image can be reduced to a size which is specified and smaller than the size of the preset background image. And then selecting a region to be fused on a preset background image, and replacing the pixel value of the pixel point corresponding to the main image in the region to be fused with the pixel value of the corresponding pixel point on the subject image. All pixel positions occupied by the main image in the fused image are reserved as the labeling data, and the fused labeled image sample is added to serve as the labeling data of the machine learning sample.

Compared with the prior art that the samples are marked in a manual mode, the method can automatically finish the image subject identification, avoid the problems of manpower consumption, low speed, insufficient response speed to new images, incapability of reaching pixel levels and the like caused by manual processing, and improve the accuracy of edge processing.

Each picture for screening the pictures to be processed in the embodiment of the invention can be a commodity display picture in the E-commerce field. Although the invention is implemented by not taking all pictures as the pictures to be processed, but screening the pictures to be processed from all the pictures, the commodity display pictures in the E-commerce field are often huge (billions or even billions of levels), even if the same commodity is sold by a plurality of E-commerce platforms, and different E-commerce platforms carry out different processing on the commodity display pictures. Even if only one of the 100 pictures is suitable, the processed labeling data of the screened to-be-processed pictures is far greater than the labeling quantity required by machine learning. In addition, the processing of the commodity display picture in the E-commerce field can be embodied at the pixel level, so that the picture boundary can be found without a large amount of operations such as gradual operation, gradient operation, local maximum operation or convolution operation, the calculated amount of image main body identification can be reduced, and the identification and labeling speed can be improved. Often, a commodity display picture mainly displays 1 to 2 commodities, so that the determination of a subject to be screened becomes extremely simple. The commodities mainly embodied in the commodity display pictures are often artificially placed in the center of the pictures, so that a main body to be expressed by the pictures can be conveniently and accurately found out.

According to a second aspect of the embodiments of the present invention, there is provided an apparatus for implementing the above method.

Fig. 7 is a schematic diagram of the main components of an apparatus for image subject recognition according to an embodiment of the present invention. As shown in fig. 7, the apparatus 700 for image subject recognition includes:

a region growing unit 701, which performs region growing on the picture to be processed, and acquires each growing region containing a non-background;

a main body determination unit 702, which sets a target region in the picture to be processed, and screens a target growth region having an intersection with the target region from each growth region; the target area is used for indicating the range of the area where the main body is located in the picture to be processed;

a subject determining unit 703, configured to determine a subject image in the to-be-processed picture according to each target growing region.

Optionally, the region growing unit or the subject determination unit is further configured to: for any pixel point, judging whether the any pixel point is a non-background pixel point or a background pixel point according to the following steps:

Optionally, the region growing unit or the main body determining unit adjusts the preset error minimum value according to the following formula:

Z′＝Z+D×Y

one or more processors;

a storage device for storing one or more programs,

Fig. 8 shows an exemplary system architecture 800 of an apparatus for image subject recognition or a method for image subject recognition to which embodiments of the present invention may be applied.

As shown in fig. 8, the system architecture 800 may include

terminal devices

801, 802, 803, a network 804, and a server 805. The network 804 serves to provide a medium for communication links between the

terminal devices

801, 802, 803 and the server 805. Network 804 may include various types of connections, such as wire, wireless communication links, or fiber optic cables, to name a few.

A user may use the

terminal devices

801, 802, 803 to interact with a server 805 over a network 804 to receive or send messages or the like. The

terminal devices

801, 802, 803 may have installed thereon various communication client applications, such as shopping-like applications, web browser applications, search-like applications, instant messaging tools, mailbox clients, social platform software, etc. (by way of example only).

The

terminal devices

801, 802, 803 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.

The server 805 may be a server that provides various services, such as a back-office management server (for example only) that supports shopping-like websites browsed by users using the

terminal devices

801, 802, 803. The backend management server may analyze and perform other processing on the received data such as the product information query request, and feed back a processing result (for example, target push information, product information — just an example) to the terminal device.

It should be noted that the method for image subject identification provided by the embodiment of the present invention is generally executed by the server 805, and accordingly, the apparatus for image subject identification is generally disposed in the server 805.

It should be understood that the number of terminal devices, networks, and servers in fig. 8 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

Referring now to FIG. 9, shown is a block diagram of a computer system 900 suitable for use with a terminal device implementing an embodiment of the present invention. The terminal device shown in fig. 9 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.

As shown in fig. 9, the computer system 900 includes a Central Processing Unit (CPU)901 that can perform various appropriate actions and processes in accordance with a program stored in a Read Only Memory (ROM)902 or a program loaded from a storage section 908 into a Random Access Memory (RAM) 903. In the RAM 903, various programs and data necessary for the operation of the system 900 are also stored. The CPU 901, ROM 902, and RAM 903 are connected to each other via a bus 904. An input/output (I/O) interface 905 is also connected to bus 904.

The following components are connected to the I/O interface 905: an input portion 906 including a keyboard, a mouse, and the like; an output section 907 including components such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 908 including a hard disk and the like; and a communication section 909 including a network interface card such as a LAN card, a modem, or the like. The communication section 909 performs communication processing via a network such as the internet. The drive 910 is also connected to the I/O interface 905 as necessary. A removable medium 911 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 910 as necessary, so that a computer program read out therefrom is mounted into the storage section 908 as necessary.

In particular, according to the embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 909, and/or installed from the removable medium 911. The above-described functions defined in the system of the present invention are executed when the computer program is executed by a Central Processing Unit (CPU) 901.

It should be noted that the computer readable medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The units described in the embodiments of the present invention may be implemented by software or hardware. The described units may also be provided in a processor, and may be described as: a processor comprising: the region growing unit is used for performing region growing on the picture to be processed and acquiring each growing region containing the non-background; the main body judging unit is used for setting a target area in the picture to be processed and screening target growing areas with intersection with the target area from each growing area; the target area is used for indicating the range of the area where the main body is located in the picture to be processed; and the main body determining unit is used for determining a main body image in the picture to be processed according to each target growing region. The names of these units do not form a limitation to the unit itself in some cases, for example, a region growing unit may also be described as a "unit that determines a subject image in the picture to be processed according to each of the target growing regions".

As another aspect, the present invention also provides a computer-readable medium that may be contained in the apparatus described in the above embodiments; or may be separate and not incorporated into the device. The computer readable medium carries one or more programs which, when executed by a device, cause the device to comprise: performing region growing on the picture to be processed, and acquiring growing regions containing non-backgrounds; setting a target area in the picture to be processed, and screening target growing areas with intersection with the target area from each growing area; the target area is used for indicating the range of the area where the main body is located in the picture to be processed; and determining a main body image in the picture to be processed according to each target growing region.

According to the technical scheme of the embodiment of the invention, the growth area is obtained through area growth, the subject image is determined according to the target growth area with intersection with the target area, the image subject identification can be automatically completed, and the problems of manpower consumption, low speed, insufficient response speed to a new image, incapability of reaching the pixel level and the like caused by manual processing are solved. By determining the target area in the picture to be processed, the main body of the picture which really needs to be highlighted can be accurately found out. The image to be processed is screened by the provided judging method, and the calculated amount of image main body identification in the subsequent steps can be reduced by adopting the provided region growing method and the region growing judging criterion, so that the accuracy of edge processing is improved. Through fusing the determined main body image with the preset background image, a labeled image sample can be obtained, a sample is provided for machine learning, and the problems that manpower is consumed, speed is low, the reaction speed to a new image is insufficient, the pixel level cannot be achieved frequently and the like caused by manually labeling the machine-learned sample are solved.

The above-described embodiments should not be construed as limiting the scope of the invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions can occur, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method of image subject recognition, comprising:

2. The method of claim 1, wherein prior to performing region growing on the picture to be processed, further comprising: screening the pictures to be processed from the pictures according to the following steps:

3. The method of claim 2, further comprising: determining the number of the frames of the preset main image, and recording as M; and each angle equally divided by M/2 at 180 degrees is used as each preset angle.

4. The method of claim 1, wherein performing region growing on the picture to be processed, and acquiring each growing region containing a non-background comprises:

5. The method of claim 1, wherein determining the subject image in the picture to be processed according to each of the target growth areas comprises:

6. The method of claim 4 or 5, further comprising: for any pixel point, judging whether the any pixel point is a non-background pixel point or a background pixel point according to the following steps:

7. The method of claim 6, wherein the preset error floor is adjusted according to the following equation:

Z′＝Z+D×Y

8. The method according to claim 1, wherein after determining the subject image in the picture to be processed according to each of the target growing regions, further comprising:

9. An apparatus for image subject recognition, comprising:

10. An electronic device for image subject recognition, comprising:

one or more processors;

a storage device for storing one or more programs,

when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-8.

11. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-8.