CN112115292A - Image search method and device, storage medium, and electronic device - Google Patents

Abstract

The invention discloses a picture searching method and device, a storage medium and an electronic device, wherein the method comprises the following steps: acquiring a first significance region of a picture to be searched, and performing hash coding on the first significance region to determine a hash coding value of the first significance region; according to the technical scheme, the problems that the calculation requirement is high, manpower and material resources are consumed and the like in the process of searching the pictures by adopting the deep learning model are solved.

Description

Image search method and device, storage medium, and electronic device

technical field

The present invention relates to the field of communications, and in particular, to a picture search method and device, a storage medium, and an electronic device.

Background technique

With the advancement of science and technology and the development of artificial intelligence, more and more intelligent algorithms are applied in daily life, especially for TV, as one of the daily household appliances with high frequency, its intelligent development is the most important The most important thing, and the most critical issue of intelligence, is to provide convenience for daily life.

In the prior art, a deep learning algorithm is used to realize image search (for example, clothing in two pictures can be detected), and according to the detection results, the target deep learning feature is extracted, and retrieved in the search database, and the one with the maximum similarity is obtained as the search result. result output. Due to the use of deep learning algorithms and cloud GPU deployment solutions, this type of algorithm requires high computing power and relies heavily on the network. In addition, it needs to collect massive data sets during training and manually label the data sets, which consumes a lot of time. Human and material resources.

In the related art, the process of using a deep learning model to search for pictures has problems such as high computing power requirements, manpower and material resources, and no effective solution has yet been proposed.

SUMMARY OF THE INVENTION

Embodiments of the present invention provide a picture search method and device, a storage medium, and an electronic device, so as to solve the problems of high computing power requirements, manpower and material resources in the process of using a deep learning model to search for pictures.

According to an optional embodiment of the present invention, a picture search method is provided, comprising: acquiring a first salient region of a picture to be searched, and performing hash coding on the first salient region to determine the first salient region. A hash code value of a saliency area; according to the hash code value, a target image matching the to-be-searched image is searched in a pre-established database, wherein a plurality of images are stored in the database.

Optionally, searching for a target picture matching the to-be-searched picture in a pre-established database according to the hash code value includes: acquiring the hash code value of the first saliency region, and storing the data in the database. A plurality of hash code values of a plurality of second salience regions, wherein, a plurality of pictures in the database respectively correspond to a plurality of second salience regions; according to the hash code values of the first salience regions The target picture is searched for a plurality of comparison results with the plurality of hash code values.

Optionally, acquiring the hash code value of the first salience area and the multiple hash code values of multiple second salience areas in the database includes: comparing the first salience area and the The second salient regions of the plurality of pictures are divided according to the same rule, and a plurality of block regions are obtained respectively; the first hash code values corresponding to the plurality of block regions of the first salience region are obtained, and the second hash code values corresponding to the plurality of block regions of the second salience region respectively.

Optionally, before searching for a target picture matching the to-be-searched picture in a pre-established database according to a comparison result between the hash code value of the first salient region and the plurality of hash code values, the The method further includes: sequentially comparing the first hash code value and the second hash code value of each second salience region to obtain a plurality of comparison results, wherein the comparison results indicate that the hash code values are the same , set the pixel value of the block area with the same hash code value as the first value, and if the comparison result indicates that the hash code value is different, set the pixel value of the block area with different hash code value as the second value .

Optionally, searching for a target picture that matches the to-be-searched picture in a pre-established database according to multiple comparison results between the hash code value of the first salient region and the plurality of hash code values, including: : determine a plurality of comparison feature maps according to the multiple comparison results, wherein the comparison feature maps are provided with values corresponding to the block regions at positions corresponding to the block regions, and the values corresponding to the block regions include at least one of the following: the first value, the second value; the target picture is determined according to the multiple comparison feature maps.

Optionally, determining the target picture according to the multiple comparison feature maps includes: determining the target comparison feature map with the largest number of first values from the multiple comparison feature maps; The picture is used as the target picture matching the to-be-searched picture.

Optionally, acquiring the first saliency area of the picture to be searched includes: preprocessing the picture to be searched; processing the preprocessed picture to be searched according to a visual attention algorithm to determine the first salient region.

According to another optional embodiment of the present invention, there is also provided a picture search apparatus, including: an acquisition module for acquiring a first saliency region of a picture to be searched; a hash coding module for The saliency area is hash-coded to determine the hash code value of the first salience area; the search module is used to search for the image matching the to-be-searched picture in the pre-established database according to the hash code value The target picture, wherein, the database stores multiple pictures.

According to yet another embodiment of the present invention, there is also provided a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, wherein the computer program is configured to execute any one of the above when running steps in a method embodiment.

According to yet another embodiment of the present invention, there is also provided an electronic device comprising a memory and a processor, wherein the memory stores a computer program, the processor is configured to run the computer program to execute any of the above Steps in Method Examples.

Through the present invention, the first salient region of the picture to be searched is obtained, and the first salient region is hash-coded to determine the hash coding value of the first salient region; according to the hash coding The value searches a pre-established database for a target picture that matches the picture to be searched. That is, first determine the saliency area of the picture, determine the hash code value of the salient area, and then find the target image that best matches the image to be searched according to the hash code, and adopt the above technical solution to solve the problem of using deep learning in the related art. The process of the model searching for pictures has problems such as high computing power requirements and labor and material resources. Furthermore, in the image search, the speed and accuracy of the search are improved without the need for large computing power requirements and without wasting a lot of manpower and material resources.

Description of drawings

The accompanying drawings described herein are used to provide further understanding of the present invention and constitute a part of the present application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the attached image:

Fig. 1 is the hardware structure block diagram of the computer terminal of a kind of picture search method of the embodiment of the present invention;

2 is a flowchart of a picture search method according to an embodiment of the present invention;

3 is a schematic flowchart of a picture search method according to an optional embodiment of the present invention;

FIG. 4 is a structural block diagram of another image search apparatus according to an embodiment of the present invention.

Detailed ways

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and in conjunction with embodiments. It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict.

It should be noted that the terms "first", "second" and the like in the description and claims of the present invention and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific sequence or sequence.

The method embodiments provided in the embodiments of the present application may be executed in a computer terminal or a similar computing device. Taking running on a computer terminal as an example, FIG. 1 is a block diagram of a hardware structure of a computer terminal of a picture search method according to an embodiment of the present invention. As shown in FIG. 1 , the computer terminal may include one or more (only one is shown in FIG. 1 ) processor 102 (the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA, etc.) As well as the memory 104 for storing data, in an exemplary embodiment, the above-mentioned computer terminal may also include a transmission device 106 and an input-output device 108 for communication functions. Those of ordinary skill in the art can understand that the structure shown in FIG. 1 is only a schematic diagram, which does not limit the structure of the above-mentioned computer terminal. For example, the computer terminal may also include more or fewer components than those shown in FIG. 1 , or have a different configuration with equivalent or more functions than those shown in FIG. 1 .

The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as computer programs corresponding to the image search method in the embodiment of the present invention. A functional application and data processing are implemented, namely, the above-mentioned method is implemented. Memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, memory 104 may further include memory located remotely from processor 102, which may be connected to a computer terminal through a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

Transmission means 106 are used to receive or transmit data via a network. The specific example of the above-mentioned network may include a wireless network provided by the communication provider of the computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC for short), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (Radio Frequency, RF for short) module, which is used to communicate with the Internet in a wireless manner.

This embodiment provides a picture search method, which is applied to the above-mentioned computer terminal. FIG. 2 is a flowchart of a picture search method according to an embodiment of the present invention, and the process includes the following steps:

Step S202, for obtaining the first saliency area of the picture to be searched;

Step S204, for performing hash coding on the first significant area to determine the hash code value of the first significant area;

Step S206 , for searching for a target picture matching the to-be-searched picture in a pre-established database according to the hash code value, wherein a plurality of pictures are stored in the database.

Through the present invention, the first salient region of the picture to be searched is obtained, and the first salient region is hash-coded to determine the hash coding value of the first salient region; according to the hash coding The value searches a pre-established database for a target picture that matches the picture to be searched. That is, first determine the saliency area of the picture, determine the hash code value of the salient area, and then find the target image that best matches the image to be searched according to the hash code, and adopt the above technical solution to solve the problem of using deep learning in the related art. The process of the model searching for pictures has problems such as high computing power requirements and labor and material resources. Furthermore, in the image search, the speed and accuracy of the search are improved without the need for large computing power requirements and without wasting a lot of manpower and material resources.

In the above step S202, there are various implementations for obtaining the first saliency region of the picture to be searched. In an optional embodiment, the picture to be searched is preprocessed; The picture to be searched is processed to determine the first saliency area of the picture to be searched.

More specifically, the preprocessing process may include performing preprocessing such as denoising and grayscale transformation on the search library image, and the technical solution for determining the salient area in the image may be implemented by a visual attention algorithm. The specific implementation process of visual attention algorithm for salient region detection includes:

1. Perform image pyramid processing on the preprocessed images to obtain 6 images of different scales;

2. Perform DCT transformation on the 6 images respectively, and normalize the transformed images with the sign function;

3. Inverse transform the processed image with Discrete Cosine Transform (referred to as DCT), and add weighted images of each scale to obtain a saliency map;

4. Use the OSTU algorithm to obtain the segmented image for the saliency map, use the connected domain method to obtain the target area, and obtain the target bounding rectangle, which is the saliency area.

It should be noted that the retrieval of the image can be any image, such as clothes, mobile phones, small bags and so on. After an image is determined, it is preprocessed first, and then DCT transform and DCT inverse transform are performed to obtain a saliency map by weighted addition of each scale image, wherein the saliency map includes saliency regions. For the saliency map, use the OSTU algorithm to obtain the segmented image, use the connected domain method to obtain the target area, and obtain the target bounding rectangle. The obtained target bounding rectangle is the saliency area detected by vision.

There may also be various ways of determining the hash code value in step S204. In an optional embodiment, the visually detected saliency region is subjected to block processing, and the size of each block is 8×8 pixels. Perform hash encoding, and the specific encoding process is as follows: The first step is to convert the RGB image into a grayscale image, and convert the image to 64-level grayscale. The second step is to calculate the average. Calculate the grayscale average of all 64 pixels. The third step is to compare the grayscale of the pixels. Compare the grayscale of each pixel with the average. If it is greater than or equal to the average value, it is recorded as 1; if it is less than the average value, it is recorded as 0. The fourth step is hash coding. The comparison results of the previous step are combined together to form a 64-bit integer. As far as the hash code value of the picture is obtained, it should be noted that the above-mentioned pixel value of 8×8 is only described as an optional implementation manner, but is not used to limit the technical solution of the embodiment of the present invention, and any pixel value can be used in the actual implementation process. .

The implementation scheme of searching for the target picture matching the to-be-searched picture in the pre-established database according to the hash code value, in an optional embodiment, it may be to obtain the hash code of the first saliency region value, and multiple hash code values of multiple second saliency regions in the database, wherein multiple pictures in the database respectively correspond to multiple second saliency regions; according to the first saliency region The target picture is searched for through the comparison results of the hash code value of the sex region and the plurality of hash code values.

That is, since there are multiple pictures stored in the database, each picture in the database is also subject to the determination of the salient region and the determination of the hash code value, that is, according to the hash code value of the image to be searched, and the value of the hash code in the database. The plurality of hash code values of the plurality of pictures searches the database for a target picture that matches the picture to be searched.

Specifically, obtaining the hash code value of the first saliency area and the multiple hash code values of multiple second salience areas in the database may also be implemented by the following manner: The saliency region and the second saliency regions of the plurality of pictures are divided according to the same rule to obtain a plurality of block regions respectively; the first hash codes corresponding to the plurality of block regions of the first salience region are obtained respectively value, and the second hash code values corresponding to the plurality of block regions of the second significant region respectively. The first saliency area and the second saliency area of the multiple pictures are divided according to the same rule, which is to obtain multiple saliency areas according to the visual attention algorithm, so that the obvious features can be retrieved in time. Find.

That is to say, through the visual attention algorithm, it is possible to obtain the hash code value of the first saliency region and the plurality of hash code values of the second saliency region of the plurality of pictures in the database, and can calculate The resources are preferentially allocated to those areas that are easy to attract the attention of the observer, which will greatly improve the work efficiency of the existing image processing and analysis methods.

It should be noted that when retrieving a picture, we search for a target picture matching the to-be-searched picture in a pre-established database according to the hash code value. That is to say, we do not directly compare the two images. For example, in traditional deep learning, we collect a large amount of data and annotations, and through the training of the model, the model is finally used to identify the image. In the image search method based on visual attention of the present invention, the feature comparison is performed through the hash code in the feature database, which is a comparison of two codes. Therefore, there is no need to collect a large amount of data and annotations, and only need to preprocess the image, process the visual attention algorithm, and extract the feature value by hash coding.

According to the comparison result of the hash code value of the first saliency region and the plurality of hash code values, before searching for the target image matching the to-be-searched image in the pre-established database, compare the first saliency area sequentially. The hash code value, and the second hash code value of each second salience region, obtain a plurality of comparison results, wherein, in the case where the comparison result indicates that the hash code value is the same, the block with the same hash code value is The pixel value of the area is set as the first value, and in the case where the comparison result indicates that the hash code value is different, the pixel value of the block area with the different hash code value is set as the second value.

That is to say, since both the first saliency region and the second saliency region are divided into regions, multiple block regions are obtained, and the hash code values between the block regions and the block regions are directly compared. If the hash codes are the same, Then, the pixel values of the partial block area are set to the first value, optionally, 255. If the hash codes are not the same, the pixel values of the partial block area are set to the second value. Optionally, If it is 0, the image with the pixel value of the first value or the second value can be understood as a comparative feature map, that is, there is a comparative feature map between the image to be searched and the image in each database, from multiple comparative feature maps. It can clearly determine the most relevant images to the images to be searched.

A plurality of comparison feature maps are determined according to the plurality of comparison results, wherein the comparison feature maps are provided with values corresponding to the block regions at positions corresponding to the block regions, and the values corresponding to the block regions include at least one of the following: the first value and the second value; determine the target picture according to the plurality of comparison feature maps, and then determine the target comparison feature map with the largest number of first values from the plurality of comparison feature maps; The picture corresponding to the target comparison feature map is used as the target picture matching the to-be-searched picture.

It can be understood that, since the saliency region in each picture is divided into multiple block regions, the comparison feature map with the largest number of pixel first values is found in the comparison feature map, so from the multiple comparison features The target contrast feature with the largest number of determined first values in the figure is both the search result.

In order to better understand the above technical solution, the optional embodiment of the present invention further provides an optional embodiment for explaining the above technical solution.

In an optional embodiment, FIG. 3 is a schematic flowchart of a picture search method according to an optional embodiment of the present invention, as shown in FIG. 3 , including the following steps:

Step S302: search for library pictures;

Step S304: perform preprocessing such as denoising and grayscale transformation on the image of the search library;

Step S306: Use the visual attention algorithm to process the pictures in the database. Perform image pyramid processing on the preprocessed images to obtain 6 images of different scales; perform DCT transformation on the 6 images respectively, and use the sign function to normalize the transformed images; 3. Inverse DCT transform, weighted addition of each scale image to obtain saliency map; use OSTU algorithm to obtain the segmented image of the saliency map, use the connected domain method to obtain the target area, and obtain the target bounding rectangle, where the obtained target bounding rectangle is significant area;

Step S308 : extracting the hash feature of the salient region, which can be understood as a process of determining the salient region and determining the hash code value. The visually detected saliency area is divided into blocks, each block is 8x8 pixels, and the divided pictures are hash-coded respectively. The specific coding process is as follows: convert the RGB image into a grayscale image, and convert the image 64 grayscale. Calculate the mean; calculate the grayscale mean of all 64 pixels. Compare the grayscale of pixels. Compare the gray level of each pixel with the average value; if it is greater than or equal to the average value, it is recorded as 1; if it is less than the average value, it is recorded as 0; hash code. Combine the comparison results of the previous step to form a 64-bit integer; get the hash code of the picture;

Step S310: Calculation in the hash index library. Compare the hash codes of the two small-block images before and after shooting, and calculate the required parameters. The specific calculation method is: if the hash codes are the same, add 1, and divide the obtained result by the length of the hash code. When the obtained parameter is greater than the set threshold, set all pixel values of the small block to 255, and set all the pixel values of the small block to 0 when it is less than the set threshold, and recombine all the small blocks into an image according to the block order, that is, get Compare the feature maps, and build a feature map retrieval database for all the images in the search database;

Step S312: extract the picture to be retrieved;

Step S314: Perform preprocessing such as denoising and grayscale transformation on the image to be retrieved;

Step S316: Visual attention algorithm. Perform image pyramid processing on the preprocessed images to obtain 6 images of different scales; perform DCT transformation on the 6 images respectively, and use the sign function to normalize the transformed images; 3. Inverse DCT transform, weighted addition of each scale image to obtain saliency map; use OSTU algorithm to obtain the segmented image of the saliency map, use the connected domain method to obtain the target area, and obtain the target bounding rectangle, where the obtained target bounding rectangle is significant area;

Step S318: Hash feature extraction. The visually detected saliency area is divided into blocks, each block is 8x8 pixels, and the divided pictures are hash-coded respectively. The specific coding process is as follows: convert the RGB image into a grayscale image, and convert the image 64 grayscale. Calculate the mean; calculate the grayscale mean of all 64 pixels. Compare the grayscale of pixels. Compare the gray level of each pixel with the average value; if it is greater than or equal to the average value, it is recorded as 1; if it is less than the average value, it is recorded as 0; hash code. Combine the comparison results of the previous step to form a 64-bit integer; get the hash code of the picture;

Step S320 : perform feature comparison through the hash code in the feature database by the hash code, and output the result.

It should be noted that the execution order of the above steps is not the actual operation flow of the solution. In the actual operation process, many steps may be operated in parallel, and the execution order of some steps may also be reversed, which is not performed in this embodiment of the present invention. limited.

According to the technical solution of the optional embodiment of the present invention, the image is first preprocessed, then the first salient region of the picture to be searched is acquired, and hash coding is performed on the first salient region to determine the first salient region. The hash code value of the saliency region; and then according to the hash code value, a target picture matching the to-be-searched picture is searched in a pre-established database. The feature comparison is performed from the feature library, and the image with the highest similarity is found as the retrieval result image. By adopting the above technical solutions, the process of using a deep learning model to search for pictures in the related art, which has high requirements on computing power and consumes manpower and material resources, is solved. Furthermore, in the image search, the speed and accuracy of the search are improved without the need for large computing power requirements and without wasting a lot of manpower and material resources.

In this embodiment, a picture search apparatus is also provided, and the apparatus is used to implement the above-mentioned embodiments and preferred implementations, and the descriptions that have been described will not be repeated. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, implementations in hardware, or a combination of software and hardware, are also possible and contemplated.

FIG. 4 is a structural block diagram of a picture search apparatus according to an embodiment of the present invention; as shown in FIG. 4 , it includes:

an obtaining module 40, configured to obtain the first saliency area of the picture to be searched;

A hash coding module 42, configured to perform hash coding on the first significant area to determine a hash code value of the first significant area;

The search module 44 is configured to search for a target picture matching the to-be-searched picture in a pre-established database according to the hash code value, wherein a plurality of pictures are stored in the database.

The acquisition module is used to acquire the first salient area of the image to be searched. The specific implementation process of the visual attention algorithm for salient area detection is as follows: 1. Perform image pyramid processing on the preprocessed image to obtain 6 images of different scales; 2. Perform DCT transformation on the 6 images respectively, and use the sign function to normalize the transformed images; 3. Use DCT inverse transformation on the processed images, and add weighted images of each scale to obtain a saliency map; 4. , Use the OSTU algorithm to obtain the segmented image for the saliency map, use the connected domain method to obtain the target area, and obtain the target bounding rectangle. The obtained target bounding rectangle is the first saliency area of the image to be searched.

A hash coding module, configured to perform hash coding on the first salience area to determine a hash code value of the first salience area. That is, hash feature extraction. The specific implementation process of the hash coding module is to perform block processing on the visually detected saliency area. Optionally, each block is 8×8 pixels in size, and hash coding is performed on the divided pictures respectively. The specific coding process is as follows: Convert the RGB image to grayscale and convert the image to 64 levels of grayscale. Calculate the mean; calculate the grayscale mean of all 64 pixels. Compare the grayscale of pixels. Compare the gray level of each pixel with the average value; if it is greater than or equal to the average value, it is recorded as 1; if it is less than the average value, it is recorded as 0; hash code. Combining the comparison results of the previous step together forms a 64-bit integer; the hash code of the picture is obtained. It should be noted that the 64-bit integer obtained by the hash code is the code for 64 pixels. If The more pixels, the more integer bits obtained by hash coding.

A search module, configured to search for a target picture matching the to-be-searched picture in a pre-established database according to the hash code value, wherein a plurality of pictures are stored in the database. The specific method of finding the module is to compare the features through the hash code in the feature library, which is the comparison of the two codes. Finally retrieved what we needed.

It should be noted that image retrieval is the retrieval of feature parts. If the retrieved image does not exist in the retrieval database, the final result will also recommend images with similar feature parts.

Through the above technical solution, the first salient region of the picture to be searched is obtained, and the first salient region is hash-coded to determine the hash coding value of the first salient region; according to the hash code The encoded value searches a pre-established database for a target picture that matches the picture to be searched. That is, first determine the saliency area of the picture, determine the hash code value of the salient area, and then find the target image that best matches the image to be searched according to the hash code, and adopt the above technical solution to solve the problem of using deep learning in the related art. The process of the model searching for pictures has problems such as high computing power requirements and labor and material resources. Furthermore, in the image search, the speed and accuracy of the search are improved without the need for large computing power requirements and without wasting a lot of manpower and material resources.

Optionally, the search module 44 is further configured to: obtain the hash code value of the first salience area and the hash code values of multiple second salience areas in the database, wherein the The plurality of pictures in the database respectively correspond to a plurality of second saliency regions; the target picture is searched according to a plurality of comparison results between the hash code value of the first salience region and the plurality of hash code values.

Optionally, the hash coding module is further configured to divide the first saliency region and the second saliency regions of the plurality of pictures according to the same rule, to obtain a plurality of block regions respectively; obtain the first saliency region; The first hash code values corresponding to the plurality of block regions of the saliency region respectively, and the second hash code values corresponding to the plurality of block regions of the second salience region respectively.

Optionally, the apparatus further includes: a processing module configured to sequentially compare the first hash code value and the second hash code value of each second significant region to obtain multiple comparison results, wherein, If the comparison result indicates that the hash code values are the same, the pixel values of the block regions with the same hash code value are set as the first value, and if the comparison result indicates that the hash code values are different, the hash code values are different The pixel value of the block region is set to the second value.

Optionally, the search module 44 is further configured to determine multiple comparison feature maps according to the multiple comparison results, wherein the comparison feature maps are provided with values corresponding to the block regions at the positions corresponding to the block regions, which are different from those of the block regions. The value corresponding to the area includes at least one of the following: the first value, the second value; and the target picture is determined according to the multiple comparison feature maps.

Optionally, the search module 44 is further configured to: determine the target comparison feature map with the largest number of first values from the multiple comparison feature maps; take the picture corresponding to the target comparison feature map as the image to be searched. Image matching target image.

Optionally, the obtaining module 40 is further configured to obtain the first saliency region of the picture to be searched, including: preprocessing the picture to be searched; processing the preprocessed picture to be searched according to a visual attention algorithm, to A first saliency region of the picture to be searched is determined.

It should be noted that the above modules can be implemented by software or hardware, and the latter can be implemented in the following ways, but not limited to this: the above modules are all located in the same processor; or, the above modules can be combined in any combination The forms are located in different processors.

An embodiment of the present invention further provides a storage medium, where a computer program is stored in the storage medium, wherein the computer program is configured to execute the steps in any one of the above method embodiments when running.

Optionally, in this embodiment, the above-mentioned storage medium may be configured to store a computer program for executing the following steps:

S1, obtaining the first saliency area of the image to be searched;

S2, performing hash coding on the first significant area to determine a hash code value of the first significant area;

S3 , searching for a target picture matching the to-be-searched picture in a pre-established database according to the hash code value, where a plurality of pictures are stored in the database.

Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (Read-Only Memory, referred to as ROM), a random access memory (Random Access Memory, referred to as RAM), Various media that can store computer programs, such as removable hard disks, magnetic disks, or optical disks.

An embodiment of the present invention also provides an electronic device, comprising a memory and a processor, where a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any of the above method embodiments.

Optionally, the above-mentioned electronic device may further include a transmission device and an input-output device, wherein the transmission device is connected to the above-mentioned processor, and the input-output device is connected to the above-mentioned processor.

Optionally, in this embodiment, the above-mentioned processor may be configured to execute the following steps through a computer program:

S1, obtaining the first saliency area of the image to be searched;

S2, performing hash coding on the first significant area to determine a hash code value of the first significant area;

S3 , searching for a target picture matching the to-be-searched picture in a pre-established database according to the hash code value, where a plurality of pictures are stored in the database.

Optionally, for specific examples in this embodiment, reference may be made to the examples described in the foregoing embodiments and optional implementation manners, and details are not described herein again in this embodiment.

Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing device, which can be centralized on a single computing device, or distributed in a network composed of multiple computing devices Alternatively, they may be implemented in program code executable by a computing device, such that they may be stored in a storage device and executed by the computing device, and in some cases, in a different order than here The steps shown or described are performed either by fabricating them separately into individual integrated circuit modules, or by fabricating multiple modules or steps of them into a single integrated circuit module. As such, the present invention is not limited to any particular combination of hardware and software.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the principle of the present invention shall be included within the protection scope of the present invention.

Claims (10)

1. an image search method, is characterized in that, comprises: Obtaining the first salient region of the picture to be searched, and performing hash coding on the first salient region to determine the hash coding value of the first salient region; According to the hash code value, a target picture matching the to-be-searched picture is searched in a pre-established database, wherein a plurality of pictures are stored in the database. 2. The method according to claim 1, wherein searching a target picture matching the picture to be searched in a pre-established database according to the hash code value, comprising: Obtain the hash code value of the first salient area and the hash code values of multiple second salience areas in the database, wherein the multiple pictures in the database correspond to multiple the second significant area; The target picture is searched according to a plurality of comparison results of the hash code value of the first saliency region and the plurality of hash code values. 3. The method according to claim 2, characterized in that, acquiring the hash code value of the first salient region, and a plurality of hash code values of a plurality of second salience regions in the database, include: dividing the first saliency area and the second saliency area of the plurality of pictures according to the same rule to obtain a plurality of block areas respectively; The first hash code values corresponding to the plurality of block regions of the first saliency region, respectively, and the second hash code values corresponding to the plurality of block regions of the second salience region, respectively, are acquired. 4. The method according to claim 3, characterized in that, according to the comparison result of the hash code value of the first salient region and the plurality of hash code values, the search is performed in a pre-established database. Before searching for a target image matching the image, the method further includes: Sequentially compare the first hash code value and the second hash code value of each second significant region to obtain a plurality of comparison results, wherein, in the case that the comparison result indicates that the hash code values are the same, the hash code value is The pixel values of the block regions with the same coding value are set as the first value, and in the case where the comparison result indicates that the hash coding values are different, the pixel values of the block regions with different hash coding values are set as the second value. 5. The method according to claim 4, characterized in that, according to a plurality of comparison results between the hash code value of the first significant region and the plurality of hash code values, a pre-established database is searched for the same value as the hash code value. The target picture matched by the picture to be searched includes: A plurality of comparison feature maps are determined according to the plurality of comparison results, wherein the comparison feature maps are provided with values corresponding to the block regions at positions corresponding to the block regions, and the values corresponding to the block regions include at least one of the following: the first value, the second value; The target picture is determined according to the plurality of comparison feature maps. 6. The method according to claim 5, wherein determining the target picture according to the plurality of contrast feature maps, comprising: determining the target contrast feature map with the largest number of first values from the plurality of contrast feature maps; The picture corresponding to the target comparison feature map is used as the target picture matching the to-be-searched picture. 7. The method according to claim 1, wherein obtaining the first saliency region of the picture to be searched comprises: Preprocessing the to-be-searched picture; The preprocessed picture to be searched is processed according to the visual attention algorithm to determine the first salient region of the picture to be searched. 8. A picture search device, characterized in that, comprising: an acquisition module, used to acquire the first saliency area of the image to be searched; a hash coding module, configured to perform hash coding on the first significant area to determine a hash code value of the first significant area; A search module, configured to search for a target picture matching the to-be-searched picture in a pre-established database according to the hash code value, wherein a plurality of pictures are stored in the database. 9. A computer-readable storage medium, wherein the computer-readable storage medium comprises a stored program, wherein the program executes the method described in any one of the preceding claims 1 to 7 when the program is run . 10. An electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute any one of claims 1 to 7 through the computer program the method described.

Images