Image output method and system for high-resolution surveillance camera

Abstract

The present invention discloses an image output method and system for a high-resolution surveillance camera, the method comprising: based on a first image captured by an image sensor at time t, synchronously acquiring a first video image and a second video image through a first image processing method and a second image processing method respectively; performing feature analysis on the acquired first video image to predict the target parameters of the first image at time t+1; the second image processing method processes the first image based on the predicted target parameters of the first image at time t to acquire the second video image. The present invention realizes dynamically updating the second image processing method while outputting the first video image based on the changed first image, acquiring the second video image output based on the dynamically updated second image processing method, and realizing simultaneous coarse-grained and fine-grained high-quality scene monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of video monitoring, and in particular to an image output method and system for a high-resolution monitoring camera. Background Art

[0002] In order to see more details, surveillance cameras have higher requirements for resolution. Most of the current surveillance cameras have 2 million to 4 million pixels. As the requirements become higher and higher, the resolution of surveillance cameras will be increased to tens of millions or even hundreds of millions of pixels. For cameras with a resolution of tens of millions of pixels, it is possible to take pictures and transmit them, but for videos, video compression and video transmission have become problems. If the resolution is retained, the compression processing speed cannot keep up, and the transmission speed also needs to be equipped with a high-bandwidth transmission medium, which is costly. Summary of the invention

[0003] In view of the problems existing in the above-mentioned prior art, the present invention provides an image output method and system for a high-resolution surveillance camera, which can simultaneously perform coarse-grained and fine-grained high-quality scene monitoring, taking into account both resolution and transmission speed. The technical solution is as follows:

[0004] In a first aspect, a method for outputting an image of a high-resolution surveillance camera is provided, comprising:

[0005] Based on the first image captured by the image sensor at time t, synchronously acquiring a first channel of video images and a second channel of video images by using a first image processing method and a second image processing method respectively, wherein the resolution of the second channel of video images is not less than that of the first image, and the resolution of the first channel of video images is not greater than that of the first image;

[0006] Based on the acquired first channel video image, a third channel video image is acquired by fusing it with the second channel video image and outputting it, and at the same time, feature analysis is performed on the acquired first channel video image to predict target parameters of the first image at time t+1;

[0007] The second image processing method processes the first image based on the predicted target parameters of the first image at time t to obtain the second channel of video images.

[0008] In a possible implementation manner, the method further includes synchronously displaying the first channel of video images and the second channel of video images through a display.

[0009] In a possible implementation, performing feature analysis on the acquired first video image to predict target parameters of the first image at time t+1 includes:

[0010] Performing feature analysis on the first video image at time t to obtain target parameters of the first video image at time t;

[0011] Based on the target parameters of the first video image at time t, predict the target parameters of the first video image at time t+1;

[0012] The target parameters of the first image at time t+1 are predicted based on the conversion relationship between the first image and the first channel of video images and the predicted target parameters of the first channel of video images at time t+1.

[0013] In a possible implementation, performing feature analysis on the first channel of video images at time t to obtain target parameters of the first channel of video images at time t includes:

[0014] Acquire a target pixel region in the first video image at time t, wherein the target pixel region is: a region formed by all pixels whose pixel change amount at the previous moment is greater than a first preset threshold and / or a pixel region whose pixel features meet the preset target features;

[0015] Based on the target pixel region of the first video image, position and size parameters of a best matching target frame for marking the position of the target pixel region are obtained.

[0016] In a possible implementation, obtaining the position and size parameters of the best matching target frame for marking the position of the target pixel area based on the target pixel area of ​​the first video image includes:

[0017] Based on the target pixel area of ​​the first video image, determining that the target in the target pixel area appears for the first time;

[0018] Based on the target pixel area of ​​the first video image, a plurality of candidate target frames are obtained;

[0019] Based on the positions and sizes of multiple candidate target frames, they are compared with preset target parameters to obtain the target frame that best matches the preset target parameters, and the position and size parameters of the best matching target frame are obtained.

[0020] Based on the target pixel area of ​​the first video image, determining that the target in the target pixel area is a non-first-appearing target;

[0021] Determine the size parameter of the best matching target frame of the target in the target pixel area based on the historical target frame size of the target in the historical video image where the target historically appeared;

[0022] The position parameters of the target best matching target box in the target pixel area are determined based on the center position of the target pixel area.

[0023] In a possible implementation, comparing the positions and sizes of the plurality of candidate target frames with preset target parameters to obtain a target frame that best matches the preset target parameters includes:

[0024] The side length and position parameters of the candidate target box are taken as a particle, and the joint function based on the proportion of non-target pixels in the candidate target box and the compatibility of the candidate target box with the area corresponding to the preset target parameters is used as the loss function to perform the particle swarm optimization algorithm.

[0025] Based on the optimal solution of the particle swarm optimization algorithm, the target frame that best matches the preset target parameters is obtained.

[0026] In a possible implementation manner, the second image processing method includes:

[0027] Based on the predicted target parameters of the first image at time t, obtaining target area cropping parameters of the first image;

[0028] The first image is cropped based on the target area cropping parameters of the first image.

[0029] In a second aspect, an image output system for a high-resolution surveillance camera is provided, comprising:

[0030] A two-channel video output unit, configured to synchronously acquire a first channel video image and a second channel video image by respectively using a first image processing method and a second image processing method based on a first image captured by an image sensor at time t, wherein the resolution of the second channel video image is not less than that of the first image, and the resolution of the first channel video image is not greater than that of the first image;

[0031] A fusion output and parameter feedback unit, used to obtain and output a third channel of video images based on the first channel of video images obtained by fusion processing with the second channel of video images, and at the same time perform feature analysis on the first channel of video images obtained to predict target parameters of the first image at time t+1;

[0032] The second image processing method processes the first image based on the predicted target parameters of the first image at time t to obtain the second channel of video images.

[0033] According to a third aspect, an electronic device is provided, the electronic device comprising:

[0034] processor;

[0035] a memory for storing processor-executable instructions;

[0036] The processor implements the image output method of a high-resolution surveillance camera as described in the first aspect by running the executable instructions.

[0037] In a fourth aspect, a computer-readable storage medium is provided, on which computer instructions are stored, and when the instructions are executed by a processor, the steps of the image output method of a high-resolution surveillance camera as described in the first aspect are implemented.

[0038] An image output method and system for a high-resolution surveillance camera of the present invention have the following beneficial effects: the high-resolution surveillance camera of the present invention can output two video streams at the same time, wherein the output first video image can perform parameter analysis to update the second image processing method, thereby outputting the first video image based on the changed first image while dynamically updating the second image processing method, obtaining the second video image output based on the dynamically updated second image processing method, wherein the resolution of the second video image is greater than the resolution of the first video image, that is, the surveillance camera of the embodiment of the present application can output two video images with different resolutions at the same time, and perform coarse-grained and fine-grained high-quality scene monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of an image output method of a high-resolution surveillance camera according to an embodiment of the present application;

[0040] Figure 2 It is a structural diagram of an image output system of a high-resolution surveillance camera according to an embodiment of the present application. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present invention. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.

[0042] The present application provides an image output method for a high-resolution surveillance camera. In one implementation, the method can be used to monitor a small number of targets in a small scene, for example, in remote indoor monitoring to monitor the movement trajectory of people indoors. The image output method for a high-resolution surveillance camera includes the following steps:

[0043] Based on the first image captured by the image sensor at time t, synchronously acquiring a first channel of video images and a second channel of video images by using a first image processing method and a second image processing method respectively, wherein the resolution of the second channel of video images is not less than that of the first image, and the resolution of the first channel of video images is not greater than that of the first image;

[0044] Based on the acquired first channel video image, a third channel video image is acquired by fusing it with the second channel video image and outputting it, and at the same time, feature analysis is performed on the acquired first channel video image to predict target parameters of the first image at time t+1;

[0045] The second image processing method processes the first image based on the predicted target parameters of the first image at time t to obtain the second channel of video images.

[0046] The high-resolution surveillance camera in the embodiment of the present application can output two video streams at the same time, wherein the output first video image can perform parameter analysis to update the second image processing method, thereby outputting the first video image based on the changed first image and dynamically updating the second image processing method, and obtaining the second video image output based on the dynamically updated second image processing method, wherein the resolution of the second video image is greater than that of the first video image, that is, the surveillance camera in the embodiment of the present application can output two video images with different resolutions at the same time, and perform coarse-grained and fine-grained high-quality scene monitoring.

[0047] Furthermore, the image output method of a high-resolution surveillance camera of an embodiment of the present application also includes synchronously displaying the first video image and the second video image through a display respectively.

[0048] In the surveillance camera of the embodiment of the present application, the FPGA side has two data transmissions, SDI and MIPI. The SDI transmission module calls the IP core inside the FPGA, and the image data is filled in accordance with the SDI standard format. SDI can select two formats, 1080p@25fps and 720p@25fps. The first video image and the second video image are output respectively through two SDI interfaces, and the two SDI interfaces can be directly connected to the display. The two video images in the MIPI data format output by the FPGA side can be connected to a dedicated video image processing module. Furthermore, the steps of fusing the first video image and the second video image to obtain the third video image and the step of performing feature analysis on the first video image to predict the target parameters of the first image at time t+1 in the embodiment of the present application can be performed in the dedicated video image processing module.

[0049] Furthermore, the above-mentioned feature analysis of the acquired first video image to predict the target parameters of the first image at time t+1 includes:

[0050] Performing feature analysis on the first video image at time t to obtain target parameters of the first video image at time t;

[0051] Based on the target parameters of the first video image at time t, predict the target parameters of the first video image at time t+1;

[0052] The target parameters of the first image at time t+1 are predicted based on the conversion relationship between the first image and the first channel of video images and the predicted target parameters of the first channel of video images at time t+1.

[0053] In an embodiment of the present application, the first image outputs a first video image through a first image processing method, and outputs a second video image through a second image processing method. Based on the target parameters of the first video image at the current moment and the next moment obtained, the target parameters of the first image can be obtained by inverse analysis of the first image processing method, and the variable parameters of the second image processing method are updated based on the target parameters of the first image at the next moment, so that the second image processing method is executed on the first image at the next moment to obtain a second video image that matches the first image at the next moment.

[0054] Furthermore, the above-mentioned feature analysis of the first video image at time t to obtain the target parameters of the first video image at time t includes:

[0055] Acquire a target pixel region in the first video image at time t, wherein the target pixel region is: a region formed by all pixels whose pixel change amount at the previous moment is greater than a first preset threshold and / or a pixel region whose pixel features meet the preset target features;

[0056] Based on the target pixel region of the first video image, position and size parameters of a best matching target frame for marking the position of the target pixel region are obtained.

[0057] The pixel feature of the pixel region meets the preset target feature, and the preset target may be a face or a vehicle. Based on the pixel region whose pixel feature meets the preset target feature, feature analysis is performed to obtain the position and size parameters of the best matching target frame of the target pixel region of the first video image at time t. The best matching target frame based on the position and size parameters can perform regional image cropping on the preset target in the first video image. Further, based on the cropped preset target region image, the second video image meeting a certain resolution requirement is obtained by processing through subsequent steps of the second image processing method. The subsequent steps of the second image processing method may include image compression, transmission, etc. For example, the image is scaled to a standard 1080p / 720p resolution for SDI output, or it can be packaged according to the MIPI protocol after image compression and output through an LVDS interface. The LVDS interface can be connected to a dedicated video image processing module. The dedicated video image processing module is used to perform feature analysis on the first video image to predict the target parameters of the first image at time t+1 and output them to the FPGA for feedback, so as to adjust the variable parameters in the second image processing method. At the same time, the first video image and the second video image can be fused in the dedicated video image processing module to obtain a third video image.

[0058] Furthermore, the above-mentioned obtaining the position and size parameters of the best matching target frame for marking the position of the target pixel area based on the target pixel area of ​​the first video image includes:

[0059] Based on the target pixel area of ​​the first video image, determining whether the target in the target pixel area is the first-appearing target;

[0060] When it is determined that the target in the target pixel area is a non-first-appearing target, size parameters of the target best matching target box in the target pixel area are determined based on the historical target box size of the target in the historical video images where the target has historically appeared, and position parameters of the target best matching target box in the target pixel area are determined based on the center position of the target pixel area.

[0061] In an embodiment of the present application, considering a target in a first video image, after the position and size parameters of the target frame are determined, the second image processing method processes the first video image based on the position and size parameters of the target frame, obtains a high-resolution image of a local area including the target, and outputs a second video image. In order to ensure the image size stability of the second video image, in continuous monitoring video image frames, for the same target, the size of its target frame is determined when it first appears, and when the target frame parameters are subsequently determined, the size of the target frame remains consistent to ensure the stability of the second video image for the same target.

[0062] When it is determined that the target in the target pixel area appears for the first time:

[0063] Based on the target pixel area of ​​the first video image, a plurality of candidate target frames are obtained;

[0064] Based on the positions and sizes of multiple candidate target frames, they are compared with preset target parameters to obtain the target frame that best matches the preset target parameters, and the position and size parameters of the best matching target frame are obtained.

[0065] Wherein, based on the target pixel area of ​​the first video image, multiple candidate target frames are obtained, including:

[0066] (1) Establish a coordinate system with a point on the edge of the target pixel area as the coordinate origin;

[0067] (2) placing a candidate box of a preset scale at the coordinate origin to obtain a candidate target box containing the target pixel;

[0068] (3) Determine whether all target pixels are within the candidate target frame;

[0069] (4) If yes, retain the position and size parameter information of the candidate target box;

[0070] (5) If not, the candidate target frame is translated and scaled based on the current coordinate origin position with a preset position movement parameter and a preset side length coordinate increment to obtain a new candidate target frame;

[0071] (6) Repeat (3)-(5) until the number of target frames retaining parameter information reaches a second preset value, and obtain a second preset value of candidate target frames.

[0072] Among them, (2) placing the candidate frame of a preset scale at the coordinate origin to obtain the candidate target frame containing the target pixel can be a point on the candidate frame of the preset scale placed at the coordinate origin, taking the position of the candidate target frame at this time as the initial position, and then translating and scaling the candidate target frame to obtain a new candidate target frame. Specifically, in an embodiment of the present application, the candidate target frame is a rectangle, a vertex of the rectangular target frame is placed at the coordinate origin, and two sides of the rectangle are placed on the horizontal axis and the vertical axis of the coordinate system respectively, and this position is used as the initial position of the candidate target frame.

[0073] Furthermore, the positions and sizes of the plurality of candidate target frames are compared with the preset target parameters to obtain the target frame that best matches the preset target parameters, including:

[0074] The side length and position parameters of the candidate target box are taken as a particle, and the joint function based on the proportion of non-target pixels in the candidate target box and the compatibility of the candidate target box with the area corresponding to the preset target parameters is used as the loss function to perform the particle swarm optimization algorithm.

[0075] Based on the optimal solution of the particle swarm optimization algorithm, the target frame that best matches the preset target parameters is obtained.

[0076] In an embodiment of the present application, when the candidate target frame is translated and scaled, in order to speed up the acquisition of the candidate target frame that can cover all target pixels and meet the best match with the preset target parameters, a particle swarm optimization algorithm is adopted, and the translation and scaling parameters are used as particles. The particle swarm optimization algorithm is performed with the joint function of the proportion of non-target pixels in the candidate target frame and the compatibility of the candidate target frame with the area corresponding to the preset target parameters as the loss function, wherein the compatibility of the candidate target frame with the area corresponding to the preset target parameters can be calculated based on the position and size parameters of the candidate target frame and the compatibility index provided by the user. For example, the compatibility index provided by the user can be the area size requirement for the target frame, the length-to-width ratio requirement for the target frame, etc. Taking the compatibility index provided by the user as the area size requirement for the target frame as an example, the loss function is calculated based on the difference between the area size of the candidate target frame and the area size of the area corresponding to the preset target parameters as part of the loss function.

[0077] Furthermore, the step of obtaining the predicted target parameter of the first video image at the next moment based on the target parameter of the first video image at the current moment includes:

[0078] (1) Obtain the position of the target pixel in the target area at the current moment;

[0079] (2) generating a plurality of candidate moving positions of the target pixel point based on the current position of the target pixel point;

[0080] (3) determining a probability value of the target pixel point moving to the candidate moving position at the next moment based on the similarity between the image features of the candidate moving position and the image features of the target pixel point position;

[0081] (4) obtaining a predicted position of the target pixel at the next moment based on the candidate moving position and the corresponding probability value, and forming a candidate target area based on the predicted positions of all target pixels at the next moment;

[0082] (5) determining whether the candidate target area meets the conditions; if not, reacquiring multiple candidate moving positions based on the current probability value distribution of the candidate moving positions, and returning to (3) until the candidate target area meets the conditions;

[0083] (6) Obtain the target pixel position of the first video image at the next moment based on the candidate target area.

[0084] Specifically, the second image processing method includes:

[0085] Based on the predicted target parameters of the first image at time t, obtaining target area cropping parameters of the first image;

[0086] The first image is cropped based on the target area cropping parameters of the first image.

[0087] In an embodiment of the present application, a first image processing method compresses and transmits the first image to output a low-resolution global surveillance video image, and a second image processing method crops a target area of ​​the first image and performs compression and transmission processing on the cropped target area, thereby realizing the transmission and display of high-resolution data for images of important areas in the scene surveillance video and retaining the detailed features of the local pixel area.

[0088] Compared with the above method embodiment, the present invention provides an image output system for a high-resolution surveillance camera, such as Figure 2 As shown, the system includes:

[0089] A two-channel video output unit, configured to synchronously acquire a first channel video image and a second channel video image by respectively using a first image processing method and a second image processing method based on a first image captured by an image sensor at time t, wherein the resolution of the second channel video image is not less than that of the first image, and the resolution of the first channel video image is not greater than that of the first image;

[0090] A fusion output and parameter feedback unit, used to obtain and output a third channel of video images based on the first channel of video images obtained by fusion processing with the second channel of video images, and at the same time perform feature analysis on the first channel of video images obtained to predict target parameters of the first image at time t+1;

[0091] The second image processing method processes the first image based on the predicted target parameters of the first image at time t to obtain the second channel of video images.

[0092] The present application also provides an electronic device, the electronic device comprising:

[0093] processor;

[0094] a memory for storing processor-executable instructions;

[0095] The processor implements the above-mentioned image output method for a high-resolution surveillance camera by running the executable instructions.

[0096] Specifically, the electronic device of the embodiment of the present application includes a processor, a memory and a communication bus, wherein the processor and the memory communicate with each other through the communication bus, and the communication bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc.

[0097] The memory may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0098] The embodiment of the present application also provides a computer-readable storage medium on which computer instructions are stored, and when the instructions are executed by a processor, the steps of the above-mentioned method for outputting images of a high-resolution surveillance camera are implemented. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)).

[0099] The present invention is not limited to the above-mentioned specific implementation modes. Various changes made by ordinary technicians in this field based on the above-mentioned concepts without creative work are all within the protection scope of the present invention.

Claims

1. A method for outputting images of a high-resolution surveillance camera, characterized in that: include: Based on the first image captured by the image sensor at time t, synchronously acquiring a first channel of video images and a second channel of video images by using a first image processing method and a second image processing method respectively, wherein the resolution of the second channel of video images is not less than that of the first image, and the resolution of the first channel of video images is not greater than that of the first image; Based on the acquired first-channel video image, a third-channel video image is obtained by fusing it with the second-channel video image and outputting it, and at the same time, a feature analysis is performed on the acquired first-channel video image to predict the target parameters of the first image at time t+1; the second image processing method processes the first image based on the predicted target parameters of the first image at time t to obtain the second-channel video image; The method of performing feature analysis on the acquired first-channel video image to predict the target parameters of the first image at time t+1 includes: performing feature analysis on the first-channel video image at time t to obtain the target parameters of the first-channel video image at time t; predicting the target parameters of the first-channel video image at time t+1 based on the target parameters of the first-channel video image at time t; predicting the target parameters of the first image at time t+1 based on the conversion relationship between the first image and the first-channel video image and the predicted target parameters of the first-channel video image at time t+1; The method of performing feature analysis on the first video image at time t to obtain target parameters of the first video image at time t includes: obtaining a target pixel area in the first video image at time t, wherein the target pixel area is an area formed by all pixels whose pixel change amount at the previous moment is greater than a first preset threshold and / or a pixel area whose pixel features meet the preset target features; and obtaining the position and size parameters of the best matching target box for marking the position of the target pixel area based on the target pixel area of ​​the first video image; The method of obtaining the position and size parameters of the best matching target frame for marking the position of the target pixel region based on the target pixel region of the first video image includes: determining that the target in the target pixel region appears for the first time based on the target pixel region of the first video image; obtaining a plurality of candidate target frames based on the target pixel region of the first video image; and obtaining a target frame that best matches the preset target parameters based on the positions and sizes of the plurality of candidate target frames, comparing them with preset target parameters, and obtaining the position and size parameters of the best matching target frame. The method compares the positions and sizes of multiple candidate target frames with preset target parameters to obtain a target frame that best matches the preset target parameters, including: taking the side length of the candidate target frame and the position parameters of the candidate target frame as a particle, and performing a particle swarm optimization algorithm based on a joint function of the proportion of the number of non-target pixels in the candidate target frame and the compatibility of the candidate target frame with the area corresponding to the preset target parameters as a loss function; and obtaining a target frame that best matches the preset target parameters based on the optimal solution of the particle swarm optimization algorithm.

2. The image output method of a high-resolution surveillance camera according to claim 1, characterized in that: The method also includes synchronously displaying the first channel video image and the second channel video image through a display.

3. The image output method of a high-resolution surveillance camera according to claim 1, characterized in that: The second image processing method comprises: Based on the predicted target parameters of the first image at time t, obtaining target area cropping parameters of the first image; The first image is cropped based on the target area cropping parameters of the first image.

4. An image output system for a high-resolution surveillance camera, characterized in that: include: A two-channel video output unit, configured to synchronously acquire a first channel video image and a second channel video image by respectively using a first image processing method and a second image processing method based on a first image captured by an image sensor at time t, wherein the resolution of the second channel video image is not less than that of the first image, and the resolution of the first channel video image is not greater than that of the first image; A fusion output and parameter feedback unit is used to obtain a third channel of video images based on the first channel of video images obtained by fusing them with the second channel of video images and output them, and at the same time, perform feature analysis on the first channel of video images obtained to predict the target parameters of the first image at time t+1; the second image processing method processes the first image based on the predicted target parameters of the first image at time t to obtain the second channel of video images; In the fusion output and parameter feedback unit, feature analysis is performed on the acquired first-channel video image to predict the target parameters of the first image at time t+1, including: feature analysis is performed on the first-channel video image at time t to obtain the target parameters of the first-channel video image at time t; based on the target parameters of the first-channel video image at time t, the target parameters of the first-channel video image at time t+1 are predicted; based on the conversion relationship between the first image and the first-channel video image and the predicted target parameters of the first-channel video image at time t+1, the target parameters of the first image at time t+1 are predicted; The method of performing feature analysis on the first video image at time t to obtain target parameters of the first video image at time t includes: obtaining a target pixel area in the first video image at time t, wherein the target pixel area is an area formed by all pixels whose pixel change amount at the previous moment is greater than a first preset threshold and / or a pixel area whose pixel features meet the preset target features; and obtaining the position and size parameters of the best matching target box for marking the position of the target pixel area based on the target pixel area of ​​the first video image; The method of obtaining the position and size parameters of the best matching target frame for marking the position of the target pixel region based on the target pixel region of the first video image includes: determining that the target in the target pixel region appears for the first time based on the target pixel region of the first video image; obtaining a plurality of candidate target frames based on the target pixel region of the first video image; and obtaining a target frame that best matches the preset target parameters based on the positions and sizes of the plurality of candidate target frames, comparing them with preset target parameters, and obtaining the position and size parameters of the best matching target frame. The method compares the positions and sizes of multiple candidate target frames with preset target parameters to obtain a target frame that best matches the preset target parameters, including: taking the side length of the candidate target frame and the position parameters of the candidate target frame as a particle, and performing a particle swarm optimization algorithm based on a joint function of the proportion of the number of non-target pixels in the candidate target frame and the compatibility of the candidate target frame with the area corresponding to the preset target parameters as a loss function; and obtaining a target frame that best matches the preset target parameters based on the optimal solution of the particle swarm optimization algorithm.

5. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing processor-executable instructions; The processor implements the method according to any one of claims 1 to 3 by running the executable instructions.

6. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the instructions are executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

Images