CN110147750B - An image search method, system and electronic device based on motion acceleration - Google Patents

Abstract

The application relates to an image searching method and system based on motion acceleration and electronic equipment. The method comprises the following steps: step a: calculating the acceleration of the target to be tracked in the current frame image according to the displacement of the first two frames of images; step b: determining a search range rectangular frame of the target to be tracked in the current frame image according to the acceleration calculation result; step c: and extracting a candidate frame of the target to be tracked in the current frame image along the diagonal line of the search range rectangular frame through an RPN (resilient packet network), and performing feature analysis on the candidate frame to obtain the position of the target to be tracked in the current frame image. According to the method, a limited search range rectangular frame is determined by an acceleration calculation mode, the candidate frame of the tracking target is selected along the diagonal line of the search range rectangular frame, a smaller search range is determined, global search is not needed, the search range is greatly reduced, the calculation amount is reduced, and the calculation speed is improved.

Description

An image search method, system and electronic device based on motion acceleration

technical field

The present application belongs to the technical field of image search, and in particular relates to an image search method, system and electronic device based on motion acceleration.

Background technique

With the development of artificial intelligence technology, more and more cutting-edge knowledge has been implemented. Among them, the object (target) tracking technology in video has received extensive attention from universities and business circles. At present, for the target tracking in the video, the general technical solution is to mark the target position to be tracked in the first frame of the video, and then perform a global search in each subsequent frame to find the target position in the next frame. target to be tracked. It is usually implemented in the following ways:

1. Searching the global image in a sliding window manner , the efficiency of this search method is relatively low, and it cannot overcome the deformation of the object during motion.

2. Using the Region Proposal Network (RPN, Region Proposal Network, Region Generation Network) for [RenS, He K, Girshick RB, et al.Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks [J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.], the advantage of this kind of network lies in the artificially set global target search method, but this method still needs to calculate a large number of images and has a wide search range.

As mentioned above, the existing image search technologies are all global searches, which will result in longer retrieval time and more computational redundancy.

SUMMARY OF THE INVENTION

The present application provides an image search method, system and electronic device based on motion acceleration, aiming to solve one of the above technical problems in the prior art at least to a certain extent.

In order to solve the above problems, the application provides the following technical solutions:

An image search method based on motion acceleration, comprising the following steps:

Step a: Calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the first two frame images;

Step b: determining the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result;

Step c: Extract the candidate frame of the target to be tracked in the current frame image through the RPN network along the diagonal of the rectangular frame of the search range, and perform feature analysis on the candidate frame to obtain the target to be tracked in the current frame image. in the location.

The technical solution adopted in the embodiment of the present application further includes: in the step a, the acceleration is a vector unit, which has both a speed and a direction, and the acceleration calculation formula is:

分别表示待追踪目标的中心位置的横、纵坐标，定义下一帧即i+2帧的起始搜索原点为：The technical solution adopted in the embodiment of the present application further includes: in the step b, determining the rectangular frame of the search range of the target to be tracked in the current frame image according to the acceleration calculation result specifically includes: The center position of the tracking target is used as the intersection of the diagonal lines of the search range rectangle, and the definition

Represent the horizontal and vertical coordinates of the center position of the target to be tracked, and define the starting search origin of the next frame, i+2 frame, as:

搜索范围矩形框的长宽分别为：Then the starting point of the search range rectangle of frame i+2 is

The length and width of the search range rectangle are:

width _i+2 =2*width _i+1 , height _i+2 =2*height _i+2 .

The technical solutions adopted in the embodiments of the present application further include: in the step c, the extraction of the candidate frame of the target to be tracked in the current frame image along the diagonal of the rectangular frame of the search range through the RPN network is specifically: Three points are obtained on the diagonal diagonal of the rectangular frame of the search range according to the preset interval distance, and then scaled again according to the set three length-width scale ratios to obtain nine candidate frames.

The technical solutions adopted in the embodiments of the present application further include: in the step a, the first two frames of images are specifically two consecutive frames of images, two frames of images at discrete intervals, or two frames of images at any time.

Another technical solution adopted by the embodiments of the present application is: an image search system based on motion acceleration, comprising:

Acceleration calculation module: used to calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the previous two frame images;

Search range calculation module: used to determine the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result;

Candidate frame extraction module: for extracting the candidate frame of the target to be tracked in the current frame image along the diagonal of the rectangular frame of the search range through the RPN network;

Target retrieval module: used to perform feature analysis on the candidate frame to obtain the position of the target to be tracked in the current frame image.

The technical solution adopted in the embodiment of the present application further includes: the acceleration is a vector unit, which has both a speed and a direction, and the acceleration calculation formula is:

分别表示待追踪目标的中心位置的横、纵坐标，定义下一帧即i+2帧的起始搜索原点为：The technical solutions adopted in the embodiments of the present application further include: the search range calculation module determining the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result specifically includes: The center position is used as the intersection of the diagonal lines of the search range rectangle, and the definition

Represent the horizontal and vertical coordinates of the center position of the target to be tracked, and define the starting search origin of the next frame, i+2 frame, as:

搜索范围矩形框的长宽分别为：Then the starting point of the search range rectangle of frame i+2 is

The length and width of the search range rectangle are:

width _i+2 =2*width _i+1 , height _i+2 =2*height _i+2 .

The technical solution adopted in the embodiment of the present application further includes: the candidate frame extraction module extracts the candidate frame of the target to be tracked in the current frame image along the diagonal of the search range rectangular frame through the RPN network. Specifically: in the search range rectangular frame Three points are obtained on the diagonal diagonal of the frame according to the preset interval distance, and then scaled again according to the three set aspect ratios to obtain nine candidate frames.

The technical solutions adopted in the embodiments of the present application further include: the first two frames of images are specifically two consecutive frames of images, two frames of images at discrete intervals, or two frames of images at any time.

Another technical solution adopted in the embodiment of the present application is: an electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,

The memory stores instructions executable by the one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the following operations of the above-described motion acceleration-based image search method :

Step a: Calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the first two frame images;

Step b: determining the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result;

Step c: Extract the candidate frame of the target to be tracked in the current frame image through the RPN network along the diagonal of the rectangular frame of the search range, and perform feature analysis on the candidate frame to obtain the target to be tracked in the current frame image. in the location.

Compared with the prior art, the beneficial effects of the embodiments of the present application are: the motion acceleration-based image search method, system, and electronic device of the embodiments of the present application use a limited search range rectangular box determined by the acceleration calculation method, and search along the The three points on the diagonal of the range rectangle frame are selected as candidate frames of the tracking target, thereby determining a smaller search range. Compared with the prior art, the present application does not need to perform global search, which greatly narrows the search range and reduces the number of searches. The amount of calculation increases the calculation speed.

Description of drawings

1 is a flowchart of an image search method based on motion acceleration according to an embodiment of the present application;

Fig. 2 (a) is the target picture of the target to be tracked in the ith frame, and Fig. 2 (b) is the target picture of the target to be tracked in the (i+1)th frame;

Figure 3 is a schematic diagram of the acceleration calculation method under the same shooting picture (the camera is fixed);

4 is a schematic diagram of a rectangular frame of the i+2 frame search range;

5 is a schematic diagram of a generation rule of an RPN network according to an embodiment of the present application;

6 is a schematic structural diagram of an image search system based on motion acceleration according to an embodiment of the present application;

FIG. 7 is a schematic structural diagram of a hardware device of an image search method based on motion acceleration provided by an embodiment of the present application.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

Please refer to FIG. 1 , which is a flowchart of an image search method based on motion acceleration according to an embodiment of the present application. The image search method based on motion acceleration according to the embodiment of the present application includes the following steps:

Step 100: mark the position of the target to be tracked in the first frame image at the beginning of the video;

Step 200: Calculate the acceleration of the target to be tracked in the current frame image and its possible direction according to the displacement of the first two frame images;

In step 200, as shown in FIG. 2, FIG. 2(a) is the target image of the target to be tracked in the ith frame, and FIG. 2(b) is the target image of the target to be tracked in the (i+1)th frame. The target tracking process in this figure is abstracted into a mathematical form, which can be represented as shown in FIG. 3 , which is an acceleration calculation method under the same shooting picture (the camera is fixed). In the embodiment of the present application, the acceleration also maintains the same properties as the acceleration in physics, and both are vector units, which have both speed and direction. The acceleration calculation formula is as follows:

It can be understood that the present application is not limited to determining the acceleration of the third frame according to the displacement of the first two consecutive frames of images, but can also use discretely spaced frames or target displacements in two frames of pictures at any time to calculate the acceleration of the current frame.

Step 300: Determine the search range rectangle frame of the target to be tracked in the current frame image according to the acceleration and direction calculation results;

分别表示待追踪目标的中心位置的横、纵坐标，定义下一帧即i+2帧的起始搜索原点为：In step 300, the calculation method of the search range rectangle is as follows: the center position of the target to be tracked in the i+1th frame is taken as the intersection of the diagonal lines of the search range rectangle, and the definition is as follows:

Represent the horizontal and vertical coordinates of the center position of the target to be tracked, and define the starting search origin of the next frame, i+2 frame, as:

搜索范围矩形框长宽分别为：Formulas (2) and (3) determine the position of the lower left origin of the starting search of the i+2 frame, then the starting point of the search range rectangle of the i+2 frame is

The length and width of the search range rectangle are:

width _i+2 = 2*width _i+1 , height _i+2 = 2*height _i+2 (4)

The above process is drawn as shown in FIG. 4 , which is a schematic diagram of a rectangular frame of the i+2 frame search range. As shown in the figure, the actual detection range of the i+2 frame is changed from the entire image of the traditional algorithm to the rectangular frame in the box in the upper right image (for clear display, the image size of the i+2 frame is enlarged, in fact, the whole The size of the image remains unchanged, only the search range rectangle is changed), thereby reducing the amount of calculation of the target search.

It can be understood that because the center of the congruent quadrilateral and the four points can determine a unique rectangular frame, the origin of the next frame can be determined whether it is through the coordinate origin or through the four boundary vertices of the previous frame.

Step 400: take three points along the diagonal of the rectangular frame of the search range according to the set interval distance through the RPN network, and extract 9 candidate frames of the target to be tracked in the current frame image according to the three aspect ratios respectively;

In step 400, the existing detection method is to first perform the operations of maintaining the original size, scaling the original image by 0.5, and expanding the original image by 2 times for each selected center position of the image in the entire image, and then in these three operations. The aspect ratios of these images are changed to 1:1, 1:2, and 2:1 respectively. Therefore, for each center point position, 3*3 candidate frames can be found for selection. There are many redundant candidate frames generated by this method. In order to save computing power and improve speed, this application no longer uses candidate frames of three scales, that is, the original size of the original image is no longer unchanged, the original image is scaled by 0.5, and the original image is scaled by 2 times. The operation of expanding the original picture, but using three points along the diagonal of the rectangular box of the search range to select the candidate box.

Step 500: Perform feature analysis on the nine candidate frames to obtain the position of the target to be tracked in the current frame image.

For details, please refer to FIG. 5 , which is a schematic diagram of a candidate frame generation rule according to an embodiment of the present application. The nine boxes in FIG. 5 are the generated candidate boxes of the target to be tracked. These nine boxes are obtained by uniformly enlarging the fixed-size candidate boxes by 1.25 times and then extracting them according to three different aspect ratios. D in Figure 5 is the diagonal diameter of the rectangular frame of the search range. Three points are obtained on the diagonal at intervals of 0.25, 0.5, and 0.75, respectively, and then 1:1 and 1:2 respectively. , and the three aspect ratios of 2:1 are rescaled to obtain nine candidate boxes. The existing RPN network detects and compares 9N candidate frames of N center points on all pictures, while this application only needs to detect 9 candidate frames after determining the search range rectangular frame, which greatly reduces the retrieval time. scope. It can be understood that parameters such as the interval distance of the points taken on the diagonal line and the scaling aspect ratio can be set according to the actual operation.

In addition, if the candidate frame exceeds the search range rectangle after scaling, the pixel value or feature value of the position of the previous frame picture is retained in this step until the selected candidate frame can accurately capture all the search range rectangles.

Please refer to FIG. 6 , which is a schematic structural diagram of an image search system based on motion acceleration according to an embodiment of the present application. The motion acceleration-based image search system in the embodiment of the present application includes a position marking module, an acceleration calculation module, a search range calculation module, a candidate frame extraction module, and a target retrieval module.

Position marking module: used to mark the position of the target to be tracked in the first frame image at the beginning of the video;

Acceleration calculation module: used to calculate the acceleration of the target to be tracked in the current frame image and its possible direction according to the displacement of the previous two frames of images; specifically, as shown in Figure 2, where Figure 2(a) is the target to be tracked. The target picture of the target in the ith frame, Figure 2(b) is the target picture of the target to be tracked in the (i+1)th frame. The target tracking process in this figure is abstracted into a mathematical form, which can be represented as shown in FIG. 3 , which is an acceleration calculation method under the same shooting picture (the camera is fixed). In the embodiment of the present application, the acceleration also maintains the same properties as the acceleration in physics, and both are vector units, which have both speed and direction. The acceleration calculation formula is as follows:

It can be understood that the present application is not limited to determining the acceleration of the third frame according to the displacement of the first two frames of the image, but can also use discretely spaced frames or target displacements in two frames of pictures at any time to perform the acceleration calculation.

分别表示待追踪目标的中心位置的横、纵坐标，定义下一帧即i+2帧的起始搜索原点为：Search range calculation module: used to determine the search range rectangle of the target to be tracked in the current frame image according to the acceleration and direction calculation results; wherein, the calculation method of the search range rectangle is: The center position is used as the intersection of the diagonal lines of the search range rectangle, and the definition

Represent the horizontal and vertical coordinates of the center position of the target to be tracked, respectively, and define the starting search origin of the next frame, i+2 frame, as:

矩形框长宽分别为：Formulas (2) and (3) determine the position of the lower left origin of the starting search of the i+2 frame, then the starting point of the search range rectangle of the i+2 frame is

The length and width of the rectangular frame are:

width _i+2 = 2*width _i+1 , height _i+2 = 2*height _i+2 (4)

The above process is drawn as shown in FIG. 4 , which is a schematic diagram of a rectangular frame of the i+2 frame search range. As shown in the figure, the actual detection range of the i+2 frame has changed from the entire picture of the traditional algorithm to the search range rectangle in the box in the upper right picture (for clear display, the size of the i+2 frame has been enlarged, the actual The size of the entire image above has remained unchanged, and only the search range rectangle has changed), thereby reducing the amount of calculation of the target search.

It can be understood that because the center of the congruent quadrilateral and the four points can determine a unique rectangular frame, the origin of the next frame can be determined whether it is through the coordinate origin or through the four boundary vertices of the previous frame.

Candidate frame extraction module: It is used to take three points along the diagonal of the rectangular frame of the search range according to the set interval distance through the RPN network, and extract 9 points of the target to be tracked in the current frame image according to the three aspect ratios respectively. Candidate frame; wherein, in order to save computing power and improve speed, this application no longer uses candidate frames of three scales, that is, the original size of the original image is no longer unchanged, the original image is scaled by 0.5, and the original image is enlarged by 2 times. Instead, three points along the diagonal of the rectangular box of the search range are used to select the candidate box.

For details, please refer to FIG. 5 , which is a schematic diagram of a generation rule of an RPN network according to an embodiment of the present application. The nine boxes in the figure are the generated candidate boxes of the target to be tracked. These nine boxes are obtained by uniformly enlarging the fixed-size candidate boxes by 1.25 times and then extracting them according to three different aspect ratios. D in Figure 4 is the diagonal diameter of the rectangular frame of the search range. Three points are obtained on the diagonal at intervals of 0.25, 0.5, and 0.75, respectively, and then 1:1 and 1:2 respectively. , and the three aspect ratios of 2:1 are rescaled to obtain nine candidate boxes. The existing RPN network detects and compares 9N candidate frames of N center points on all pictures, while this application only needs to detect 9 candidate frames after determining the search range rectangular frame, which greatly reduces the retrieval time. scope. It can be understood that parameters such as the interval distance of the points taken on the diagonal line and the scaling aspect ratio can be set according to the actual operation.

In addition, if the candidate frame exceeds the search range rectangle after scaling, the pixel value or feature value of the position of the previous frame picture is retained in this step until the selected candidate frame can accurately capture all the search range rectangles.

Target retrieval module: It is used to perform feature analysis on 9 candidate frames to obtain the position of the target to be tracked in the current frame image.

FIG. 7 is a schematic structural diagram of a hardware device of an image search method based on motion acceleration provided by an embodiment of the present application. As shown in Figure 7, the device includes one or more processors and memory. Taking a processor as an example, the device may further include: an input system and an output system.

The processor, the memory, the input system and the output system may be connected by a bus or in other ways, and the connection by a bus is taken as an example in FIG. 7 .

As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules. The processor executes various functional applications and data processing of the electronic device by running the non-transitory software programs, instructions and modules stored in the memory, that is, the processing method of the above method embodiment is implemented.

The memory may include a stored program area and a stored data area, wherein the stored program area can store an operating system and an application program required by at least one function; the stored data area can store data and the like. Additionally, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory may optionally include memory located remotely from the processor, which may be connected to the processing system via 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.

The input system can receive input numerical or character information and generate signal input. The output system may include a display device such as a display screen.

The one or more modules are stored in the memory, and when executed by the one or more processors, perform the following operations of any of the foregoing method embodiments:

Step a: Calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the first two frame images;

Step b: determining the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result;

Step c: Extract the candidate frame of the target to be tracked in the current frame image through the RPN network along the diagonal of the rectangular frame of the search range, and perform feature analysis on the candidate frame to obtain the target to be tracked in the current frame image. in the location.

The above product can execute the method provided by the embodiments of the present application, and has functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, reference may be made to the method provided in this embodiment of the present application.

An embodiment of the present application provides a non-transitory (non-volatile) computer storage medium, where the computer storage medium stores computer-executable instructions, and the computer-executable instructions can perform the following operations:

Step a: Calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the first two frame images;

Step b: determining the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result;

Step c: Extract the candidate frame of the target to be tracked in the current frame image through the RPN network along the diagonal of the rectangular frame of the search range, and perform feature analysis on the candidate frame to obtain the target to be tracked in the current frame image. in the location.

An embodiment of the present application provides a computer program product, the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by a computer , which causes the computer to do the following:

Step a: Calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the first two frame images;

Step b: determining the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result;

Step c: Extract the candidate frame of the target to be tracked in the current frame image through the RPN network along the diagonal of the rectangular frame of the search range, and perform feature analysis on the candidate frame to obtain the target to be tracked in the current frame image. in the location.

The motion acceleration-based image search method, system, and electronic device of the embodiments of the present application use an acceleration calculation method to determine a limited search range rectangular frame, and track three points along the diagonal of the search range rectangular frame to track the candidate frame of the target Therefore, a smaller search range is determined. Compared with the prior art, the present application does not need to perform a global search, which greatly reduces the search range, reduces the amount of calculation, and improves the calculation speed.

The above description of the disclosed embodiments enables any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined in this application may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (7)

1. an image search method based on motion acceleration, is characterized in that, comprises the following steps: Step a: Calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the first two frame images; Step b: determining the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result; Step c: Extract the candidate frame of the target to be tracked in the current frame image through the RPN network along the diagonal of the rectangular frame of the search range, and perform feature analysis on the candidate frame to obtain the target to be tracked in the current frame image. position in; In the step a, the acceleration is a vector unit, which has both magnitude and direction, and the acceleration expression is:

In the step c, the extracting the candidate frame of the target to be tracked in the current frame image along the diagonal of the search range rectangular frame by the RPN network is specifically: on the diagonal diagonal line of the search range rectangular frame, respectively Three points are obtained according to the preset interval distance, and then scaled again according to the three set aspect ratios to obtain nine candidate frames; the three points are respectively located at 0.25 times the length of the diagonal line , 0.5 times and 0.75 times; the three length and width scales are: 1:1, 1:2 and 2:1 respectively. 2 . The image search method based on motion acceleration according to claim 1 , wherein in the step b, the rectangular frame for determining the search range of the target to be tracked in the current frame image according to the acceleration calculation result specifically includes: 2 . : Take the center position of the target to be tracked in the i+1th frame image as the intersection of the diagonal lines of the search range rectangle, define

Represent the horizontal and vertical coordinates of the center position of the target to be tracked, and define the starting search origin of the next frame, i+2 frame, as:

Then the starting point of the search range rectangle of frame i+2 is

The length and width of the search range rectangle are: width _i+2 =2*width _i+1 , height _i+2 =2*height _i+1 . 3. The image search method based on motion acceleration according to any one of claims 1 to 2, wherein in the step a, the first two frames of images are specifically two continuous frames of images or two discretely spaced images. frame image. 4. an image search system based on motion acceleration, is characterized in that, comprises: Acceleration calculation module: used to calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the previous two frame images; Search range calculation module: used to determine the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result; Candidate frame extraction module: for extracting the candidate frame of the target to be tracked in the current frame image along the diagonal of the rectangular frame of the search range through the RPN network; Target retrieval module: used to perform feature analysis on the candidate frame to obtain the position of the target to be tracked in the current frame image; The acceleration is a vector unit, which has both magnitude and direction, and the acceleration expression is:

The candidate frame extracting module extracts the candidate frame of the target to be tracked in the current frame image along the diagonal of the search range rectangular frame through the RPN network, specifically: on the diagonal diagonal of the search range rectangular frame respectively according to preset Obtain three points at the interval distance of , and then scale them again according to the set three aspect ratios to obtain nine candidate frames; the three points are located at 0.25 times the length of the diagonal diagonal, 0.5 times and 0.75 times; the three length-width scales are: 1:1, 1:2 and 2:1, respectively. 5. The image search system based on motion acceleration according to claim 4, wherein the search range calculation module determines the search range rectangle frame of the target to be tracked in the current frame image according to the acceleration calculation result specifically comprising: The center position of the target to be tracked in the i+1 frame image is taken as the intersection of the diagonal lines of the rectangular frame of the search range, which defines

Represent the horizontal and vertical coordinates of the center position of the target to be tracked, respectively, and define the starting search origin of the next frame, i+2 frame, as:

Then the starting point of the search range rectangle of frame i+2 is

The length and width of the search range rectangle are: width _i+2 =2*width _i+1 , height _i+2 =2*height _i+1 . 6 . The motion acceleration-based image search system according to claim 4 , wherein the first two frames of images are specifically two consecutive frames of images or two frames of images at discrete intervals. 7 . 7. An electronic device comprising: at least one processor; and a memory communicatively coupled to the at least one processor; wherein, The memory stores instructions executable by the one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the motion-based motion described in any of 1 to 3 above The following operations of the image search method of acceleration: Step a: Calculate the acceleration of the target to be tracked in the current frame image according to the displacement of the first two frame images; Step b: determining the search range rectangle of the target to be tracked in the current frame image according to the acceleration calculation result; Step c: Extract the candidate frame of the target to be tracked in the current frame image through the RPN network along the diagonal of the rectangular frame of the search range, and perform feature analysis on the candidate frame to obtain the target to be tracked in the current frame image. position in; In the step a, the acceleration is a vector unit, which has both magnitude and direction, and the acceleration expression is:

In the step c, the extraction of the candidate frame of the target to be tracked in the current frame image along the diagonal of the search range rectangular frame through the RPN network is specifically: on the diagonal diagonal of the search range rectangular frame, respectively Three points are obtained according to the preset interval distance, and then scaled again according to the set three aspect ratios to obtain nine candidate frames; the three points are respectively located at 0.25 times the length of the diagonal diagonal. , 0.5 times and 0.75 times; the three length and width scales are: 1:1, 1:2 and 2:1 respectively.

Images