Multi-object Trajectory Generation Method, System, Electronic Device, and Storage Medium

Abstract

The present invention relates to the field of video tracking technology, and particularly to a method, a system, an electronic device and a storage medium for generating multi-target object trajectories. The method includes: obtaining video data containing multi-target objects; performing frame division processing on the obtained video data, selecting a blank video frame as the background, determining the target objects to be tracked in the video frames after frame division, and generating position frames for the target objects to be tracked; labeling the corresponding position frames according to the tracking feature points of the target objects to be tracked, and calculating the center points of each position frame; obtaining the center point coordinates corresponding to the multi-target objects in consecutive video frames, mapping them to the corresponding video data according to the time sequence, and generating the movement trajectory results of the multi-target objects in the video data. The present invention solves the problem of difficult multi-target tracking and generating movement trajectories, generates continuous center point trajectories in the video, and avoids the situation of losing multi-target objects or incorrect trajectories.

Description

Technical Field

[0001] The present invention relates to the technical field of video tracking, and particularly to a method, a system, an electronic device, and a storage medium for generating multi-target object trajectories. Background Art

[0002] With the continuous advancement of the construction of smart cities and the continuous upgrading of security monitoring, a large-scale video monitoring network at the national level has initially taken shape. At the same time, one of the most direct consequences of large-scale surveillance videos is the generation of a vast amount of video data. The vast amount of video data poses a huge challenge to traditional intelligent video analysis techniques.

[0003] When analyzing and processing the videos collected by surveillance cameras, one of the very important tasks is to capture and track multiple targets in the dynamically captured surveillance videos. For the tracking of each target, the current processing method is single-target tracking technology, which is relatively simple. When the number of processing targets increases, it needs to be processed separately, increasing the difficulty of processing. Especially when it is necessary to generate corresponding moving trajectories for multi-target objects, the existing processing algorithms cannot meet the actual application requirements. Summary of the Invention

[0004] To solve the problem of tracking multi-target objects and generating corresponding moving trajectories, the present invention provides a method, a system, an electronic device, and a storage medium for generating multi-target object trajectories. Based on the recognition and processing of multiple target sample images, video tracking is performed on the recognized multiple target images, and the moving trajectory results of multi-target objects are output.

[0005] To achieve the above object, the embodiments of the present invention provide the following technical solutions:

[0006] In a first aspect, in an embodiment provided by the present invention, a method for generating multi-target object trajectories is provided, including:

[0007] Obtain video data containing multi-target objects;

[0008] Perform frame splitting on the obtained video data, select a blank video frame as the background, determine the target objects to be tracked in the video frame after frame splitting, and generate position boxes for the target objects to be tracked;

[0009] Label the corresponding position boxes according to the tracking feature points of the target objects to be tracked, and calculate the center points of each position box;

[0010] Obtain the center point coordinates corresponding to the multi-target objects in consecutive video frames, map them to the corresponding video data according to the time sequence, and generate the moving trajectory results of the multi-target objects in the video data.

[0011] In some embodiments provided by the present invention, determining the target object to be tracked includes:

[0012] Selecting a blank video frame as the background and comparing it with video frames one by one;

[0013] Using the background subtraction method to identify the foreground area in the current video frame to obtain the foreground image block in the current video frame;

[0014] Dividing the foreground image block to determine multiple target objects to be tracked.

[0015] In some embodiments provided by the present invention, determining multiple target objects to be tracked includes:

[0016] Obtaining the foreground image block in the current video frame;

[0017] Using an image recognition algorithm to identify local feature points in the foreground image block;

[0018] Obtaining the difference result of the global feature in the foreground image block according to the local feature points, and dividing the foreground image block according to the difference result to obtain multiple target objects to be tracked.

[0019] In some embodiments provided by the present invention, the image recognition algorithm for local feature points includes: blob detection algorithm and corner detection algorithm. The blob detection algorithm includes Laplacian of Gaussian operator detection (LOG), pixel Hessian matrix, and determinant of Hessian (DOH) detection; the corner detection algorithm includes Harris corner feature extraction and FAST corner feature extraction.

[0020] In some embodiments provided by the present invention, generating the position box of the target object to be tracked is to generate a contour with color and gray-scale differences between the foreground image block of the target object to be tracked and the surrounding area according to the two types of local feature points of blobs and corners, and the generated contour forms the position box.

[0021] In some embodiments provided by the present invention, it further includes:

[0022] Segmenting the foreground image blocks of each of the multiple target objects to be tracked obtained, dividing the foreground image block into multiple small blocks, and calculating the histogram value and LBP feature value of each small block to form a high-dimensional vector representing the small block;

[0023] Capturing the target object to be tracked according to the high-dimensional vectors of the multiple small blocks.

[0024] In some embodiments provided by the present invention, according to the different high-dimensional vectors corresponding to the image blocks where multiple target objects are located, numbering the position boxes formed by the generated contours, and the numbers of the position boxes corresponding to the same target object are the same in consecutive video frames.

[0025] In some embodiments provided by the present invention, the method for calculating the center points of each position box includes:

[0026] Generate the contour of the target object to be tracked according to the spots and corner points;

[0027] Select scattered points from the generated contour and draw the diagonal of the scattered point graph to determine the center point of the corresponding position box of the target object to be tracked.

[0028] In a second aspect, in another embodiment provided by the present invention, a multi-target object trajectory generation system is provided. The multi-target object trajectory generation system uses the above multi-target object trajectory generation method to generate multi-target object movement trajectories in video data. The multi-target object trajectory generation system includes a target object to be tracked recognition module, a framing module, a center point calculation module, and a trajectory generation module.

[0029] The target object to be tracked recognition module is used to perform frame-by-frame processing on the acquired video data, select a blank video frame as the background, and use the background subtraction method to determine the target object to be tracked in the video frame after frame division.

[0030] The framing module is used to generate the contour of the target object to be tracked according to local feature points, and the generated contour forms a position box to obtain the position box of the target object to be tracked.

[0031] The center point calculation module is used to label the corresponding position boxes of the tracking feature points of the target object to be tracked, select scattered points on the contour to form a geometric figure, and calculate the center points of the geometric figures corresponding to each position box; and

[0032] The trajectory generation module is used to map the center point coordinates of the multi-target objects in the acquired consecutive video frames to the corresponding video data according to the time sequence to generate the multi-target object movement trajectory result in the video data.

[0033] In a third aspect, in another embodiment provided by the present invention, an electronic device is provided, including a memory and a processor. The memory stores a computer program, and when the processor loads and executes the computer program, the steps of the multi-target object trajectory generation method are implemented.

[0034] In a fourth aspect, in another embodiment provided by the present invention, a storage medium is provided, storing a computer program, and when the computer program is loaded and executed by a processor, the steps of the multi-target object trajectory generation method are implemented.

[0035] The technical solution provided by the present invention has the following beneficial effects:

[0036] The multi-target object trajectory generation method, system, electronic device, and storage medium provided by the present invention are directed to provided video data. Taking a specified blank video frame as a reference frame, determining multi-target objects in each video frame and configuring position boxes, determining the center points and coordinates of the multi-target objects, mapping them to the video data to form continuous frame center point markings, and forming multi-target object movement trajectory results in the video data, which can effectively solve the problem of difficult multi-target tracking and generating movement trajectories. When multi-target objects are recognized and tracked, continuous center point trajectories will be generated in the video, avoiding the situation of losing multi-target objects or incorrect trajectories, and improving the accuracy of multi-target object trajectory generation.

[0037] These aspects or other aspects of the present invention will be more clearly understood in the following description of the embodiments. It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present invention. In the drawings:

[0039] Figure 1 It is a flowchart of a multi-target object trajectory generation method according to an embodiment of the present invention.

[0040] Figure 2 It is a flowchart of determining a target object to be tracked in a multi-target object trajectory generation method according to an embodiment of the present invention.

[0041] Figure 3 It is a flowchart of determining multiple target objects to be tracked in a multi-target object trajectory generation method according to an embodiment of the present invention.

[0042] Figure 4 It is a flowchart of calculating the center point of a position box in a multi-target object trajectory generation method according to an embodiment of the present invention.

[0043] Figure 5 It is a schematic diagram of determining the geometric center point of a target object in a multi-target object trajectory generation method according to an embodiment of the present invention.

[0044] Figure 6 It is a schematic diagram of determining the geometric center point of another target object in a multi-target object trajectory generation method according to an embodiment of the present invention.

[0045] Figure 7 It is a system block diagram of a multi-target object trajectory generation system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in 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 invention and are not used to limit the present invention.

[0047] In some processes described in the specification, claims and above-mentioned drawings of the present invention, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0048] Next, the technical solutions in the exemplary embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the exemplary embodiments of the present invention. Obviously, the described exemplary embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0049] Since the technology for capturing and tracking targets in videos in existing applications usually performs single-target tracking, this method is relatively simple. When the number of targets to be processed increases, separate processing is required, which increases the difficulty of processing. As a result, it is impossible to generate corresponding movement trajectories for multiple target objects in a video.

[0050] To address the above problems, the present invention provides a method, system, electronic device, and storage medium for generating multi-target object trajectories, which perform video tracking on multiple identified target images and output the results of multi-target object movement trajectories.

[0051] Specifically, the embodiments of the present application will be further described below with reference to the accompanying drawings.

[0052] As Figure 1 shown, an embodiment of the present invention provides a method for generating multi-target object trajectories, including the following steps:

[0053] S1. Obtain video data containing multi-target objects, perform frame division processing on the obtained video data, select a blank video frame as the background, determine the target objects to be tracked in the video frame after frame division, and generate position frames for the target objects to be tracked.

[0054] S2. Number the corresponding position boxes according to the tracking feature points of the target object to be tracked, and calculate the center points of each position box.

[0055] S3. Obtain the center point coordinates corresponding to multiple target objects in consecutive video frames, map them to the corresponding video data according to the time sequence, and generate the moving trajectory results of multiple target objects in the video data.

[0056] In this embodiment, the consecutive frames of video data are obtained in a frame-by-frame manner. The video frame without any target object in the captured image is used as the background and compared with other video frames to determine the target objects to be tracked in other video frames. Then, the center points of each target object to be tracked are determined, and the center point coordinates of each target object to be tracked in consecutive video frames are mapped to the video data to form the moving trajectory results of multiple target objects in the video data.

[0057] During processing, after selecting a blank video frame, during the video data acquisition process, the currently captured video frame can be compared to determine each target object to be tracked in the current video frame, calculate the center point position and coordinates, and form the moving trajectory results of consecutive frames in the video data with the center points of the consecutive frames before the current video frame, which is efficient and convenient to process.

[0058] In step S1 of the present invention, refer to Figure 2 as shown, the method for determining the target object to be tracked is:

[0059] S101. Select a blank video frame as the background and compare it with video frames one by one;

[0060] S102. Use the background subtraction method to identify the foreground area in the current video frame to obtain the foreground image block in the current video frame;

[0061] S103. Divide the foreground image block to determine multiple target objects to be tracked.

[0062] In this embodiment, when using the background subtraction method for foreground area identification, a blank video frame is selected as the reference frame and compared with other video frames one by one to distinguish the foreground and background of the video frame image. The selected reference frame is used for foreground area detection to identify the foreground area in other video frame images. In the embodiment of the present invention, the background subtraction method is used to detect the foreground area of the image and perform morphological processing to obtain the foreground image block.

[0063] Preferably, as an implementable manner, refer to Figure 3 as shown, the steps for determining multiple target objects to be tracked in step S103 include the following steps:

[0064] S1031. Obtain the foreground image block in the current video frame;

[0065] S1032. Identify local feature points in the foreground image block using an image recognition algorithm;

[0066] S1033. Obtain the difference result of the global feature in the foreground image block according to the local feature points, and divide the foreground image block according to the difference result to obtain multiple targets to be tracked.

[0067] In this embodiment, local feature points in the foreground image block are identified, and the foreground image block is divided according to different local feature points, so as to obtain multiple targets to be tracked. The specific principle is to identify local feature points through blobs and corners. It has good stability and is not easily interfered by the external environment when identifying the feature points of the current frame.

[0068] In the embodiment of the present invention, the local feature points may include color features, texture features, shape features, and local feature points, etc. For these two types of local feature points, namely blobs and corners. A blob usually refers to an area with a color and grayscale difference from the surrounding area, such as a pedestrian or a vehicle on the road. It is an area, so it has stronger noise resistance and better stability than corners. A corner is the intersection of the corners of an object or the intersection of lines in an image.

[0069] When identifying local feature points of blobs and corners, the image recognition algorithm includes a blob detection algorithm and a corner detection algorithm. The blob detection algorithm includes Laplacian of Gaussian operator detection (LOG), pixel Hessian matrix, and determinant of Hessian (DOH) detection; the corner detection algorithm includes Harris corner feature extraction and FAST corner feature extraction.

[0070] Preferably, as an implementable embodiment, when detecting spots, the LOG and DOH algorithms are used. The methods for spot detection mainly include the method of detecting using the Laplacian of Gaussian operator (LOG), and the method of using the Hessian matrix (second-order differential) of pixel points and its determinant value (DOH). The LoG method is to detect image spots using the Laplace of Gaussian (LOG) operator. During detection, the convolution operation of an image with a two-dimensional function is actually to find the similarity between the image and this function. Similarly, the convolution of an image with the Laplace of Gaussian function is actually to find the similarity between the image and the Laplace of Gaussian function. When the size of the spots in the image approaches the shape of the Laplace of Gaussian function, the Laplacian response of the image reaches the maximum. Since the Laplacian kernel of the two-dimensional Gaussian function is very similar to a spot, convolution can be used to find the spot-like structures in the image. The DoH method is to use the second-order differential Hessian matrix of image points and the value of the determinant of the Hessian matrix DoH (Determinant of Hessian). The value of the determinant of the Hessian matrix also reflects the local structural information of the image. Compared with LoG, DoH has a better inhibitory effect on the spots with slender structures in the image.

[0071] Preferably, as an implementable embodiment, the position box of the target object to be tracked is generated by generating a foreground image block of the target object to be tracked and a contour with color and gray differences around it according to the spots and corner points, which are two types of local feature points, and the generated contour forms the position box.

[0072] As an implementable embodiment of the present invention, the multi-target object trajectory generation method further includes:

[0073] Segment the foreground image blocks of each of the multiple targets to be tracked obtained, divide the foreground image blocks into multiple small blocks, and calculate the histogram value and LBP feature value of each small block to form a high-dimensional vector representing the small block;

[0074] Capture the targets to be tracked according to the high-dimensional vectors of the multiple small blocks.

[0075] In this embodiment, after separating the foreground image blocks of each of the multiple targets to be tracked, each foreground image block is further divided into multiple small blocks, and the histogram value and LBP feature value of each small block are calculated to form a high-dimensional vector representing the small block, so as to improve the tracking accuracy and recognition efficiency of identifying multiple targets to be tracked.

[0076] Preferably, as an implementable embodiment, in step S2, as shown in Figure 4 The method for calculating the center points of each position box includes the following steps:

[0077] S201. Generate the contour of the target object to be tracked based on the spots and corner points;

[0078] S201. Select scattered points from the generated contour, draw the diagonal of the scattered point graph, and determine the center point of the corresponding position box of the target object to be tracked.

[0079] In this embodiment, the scattered points are selected on the generated contour according to a set number, the step length between adjacent scattered points is equal, and a graph of selected scattered points is formed along the contour to determine the geometric center point and coordinates of the scattered point graph. Among them, obtain the contour length of each target object and mark the selected scattered points as scattered points according to the equal step length, and then use the scattered points as the vertices of the geometric figure. Optionally, as an implementable manner, see Figure 5 and Figure 6 As shown, the set number of scattered points is 4. Take the four selected scattered points as vertices to form a quadrilateral, connect the diagonal scattered points, and the intersection position is used as the geometric center point of the target object, and mark the coordinates corresponding to the geometric center point.

[0080] In step S3, the geometric center points obtained from each video frame are mapped to the video data according to their coordinates, and the center points in consecutive frames form the moving trajectory results corresponding to each target object in the video. Multiple targets can be tracked and distinguished by the labels of the position boxes.

[0081] The method of the present invention, for the provided video data, takes the specified blank video frame as the reference frame, determines the multi-target objects of each video frame and configures the position boxes, determines the center points and coordinates of the multi-target objects, maps them to the video data to form the consecutive frame center point marks, and forms the multi-target object moving trajectory results in the video data. It can effectively solve the problem of difficult multi-target tracking and generating motion trajectories. When the multi-target objects are recognized and tracked, continuous center point trajectories will be generated in the video, avoiding the situation of losing multi-target objects or incorrect trajectories, and improving the accuracy of multi-target object trajectory generation.

[0082] In an embodiment of the present invention, as shown in Figure 6 The present invention also discloses a multi-target object trajectory generation system. The multi-target object trajectory generation system uses the above multi-target object trajectory generation method to generate multi-target object moving trajectories in the video data; the multi-target object trajectory generation system includes a target object to be tracked recognition module 100, a framing module 200, a center point calculation module 300, and a trajectory generation module 400.

[0083] The target object to be tracked recognition module 100 is used to perform frame-by-frame processing on the acquired video data, select the blank video frame image as the background, and use the background subtraction method to determine the target object to be tracked in the video frame image after frame division.

[0084] After selecting a blank video frame, during the video data acquisition process, the currently acquired video frame can be compared to determine each target object to be tracked in the current video frame, calculate the center point position and coordinates, and form the moving trajectory result of consecutive frames in the video data with the center points of the consecutive frames before the current video frame.

[0085] In this embodiment, when the target object to be tracked recognition module 100 determines the target object to be tracked, a blank video frame is selected as the background and compared with the video frames one by one; the foreground region in the current video frame is recognized by using the background subtraction method to obtain the foreground image block in the current video frame; the foreground image block is divided to determine multiple targets to be tracked.

[0086] In this embodiment, when the target object to be tracked recognition module 100 determines multiple targets to be tracked, the foreground image block in the current video frame is obtained; the local feature points in the foreground image block are recognized by using an image recognition algorithm; according to the local feature points, the difference result of the global features in the foreground image block is obtained, and the foreground image block is divided according to the difference result to obtain multiple targets to be tracked.

[0087] For the foreground image block, the local feature points therein are recognized, and the foreground image block is divided according to the differences of the local feature points, so as to obtain multiple targets to be tracked. The specific principle is to recognize the local feature points through blobs and corners. It has good stability and is not easily interfered by the external environment when recognizing the feature points of the current frame.

[0088] The framing module 200 is used to generate the contour of the target object to be tracked according to the local feature points, and the generated contour forms a position box to obtain the position box of the target object to be tracked.

[0089] By using blobs and corners as local feature points for recognition, the local feature points are recognized by using the blob detection algorithm and the corner detection algorithm included in the image recognition algorithm. The blob detection algorithm includes the Laplacian of Gaussian operator detection (LOG), the pixel Hessian matrix, and the determinant of Hessian (DOH) detection; the corner detection algorithm includes the Harris corner feature extraction and the FAST corner feature extraction.

[0090] When performing blob detection, the LOG and DOH algorithms are used. The methods of blob detection mainly include the method of using the Laplacian of Gaussian operator detection (LOG) and the method of using the pixel Hessian matrix (second-order differential) and its determinant value (DOH). The method of LoG is to detect image blobs by using the Laplace of Gaussian (LOG) operator.

[0091] The framed module 200 generates a position box for the target object to be tracked. Based on the fact that spots and corner points are two types of local feature points, it generates a foreground image block of the target object to be tracked and a contour with color and grayscale differences around it, and the generated contour forms the position box.

[0092] The center point calculation module 300 is used to label the corresponding position boxes for the tracking feature points of the target object to be tracked, select scattered points on the contour to form a geometric figure, and calculate the center points of the geometric figures corresponding to each position box.

[0093] In this embodiment, when the center point calculation module 300 calculates the center points of each position box, it generates a contour of the target object to be tracked according to spots and corner points; selects scattered points from the generated contour, and draws the diagonal of the scattered point figure to determine the center points of the corresponding position boxes of the target object to be tracked.

[0094] Optionally, the scattered points are selected on the generated contour according to a set number, the step size between adjacent scattered points is equal, a figure of selected scattered points is formed along the contour, and the geometric center point and coordinates of the scattered point figure are determined.

[0095] The trajectory generation module 400 is used to map the center point coordinates corresponding to multiple target objects in the acquired consecutive video frames to the corresponding video data according to the time sequence, and generate the moving trajectory results of the multiple target objects in the video data.

[0096] In this embodiment, the geometric center points obtained for each video frame are mapped to the video data according to their coordinates, and the center points in consecutive frames form the moving trajectory results corresponding to each target object in the video. Multiple targets can be tracked and distinguished by the labels of the position boxes.

[0097] It should be particularly noted that the multi-target object trajectory generation system executes using the steps of a multi-target object trajectory generation method as described above. Therefore, the operation process of the multi-target object trajectory generation system in this embodiment will not be introduced in detail.

[0098] In one embodiment, in the embodiments of the present invention, an electronic device is further provided, including at least one processor, and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor. When the instructions are executed by the at least one processor, the at least one processor executes the multi-target object trajectory generation method, and when the processor executes the instructions, it implements the steps in the above method embodiments:

[0099] Obtain video data containing multiple target objects;

[0100] Perform frame segmentation on the acquired video data, select a blank video frame as the background, determine the target object to be tracked in the video frame after segmentation, and generate a position box for the target object to be tracked.

[0101] Number the corresponding position boxes according to the tracking feature points of the target object to be tracked, and calculate the center points of each position box.

[0102] Obtain the center point coordinates corresponding to multiple target objects in consecutive video frames, map them to the corresponding video data according to the time sequence, and generate the moving trajectory results of multiple target objects in the video data.

[0103] In an embodiment of the present invention, an electronic device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented:

[0104] Obtain video data containing multiple target objects;

[0105] Perform frame segmentation on the acquired video data, select a blank video frame as the background, determine the target object to be tracked in the video frame after segmentation, and generate a position box for the target object to be tracked.

[0106] Number the corresponding position boxes according to the tracking feature points of the target object to be tracked, and calculate the center points of each position box.

[0107] Obtain the center point coordinates corresponding to multiple target objects in consecutive video frames, map them to the corresponding video data according to the time sequence, and generate the moving trajectory results of multiple target objects in the video data.

[0108] In an embodiment of the present invention, a storage medium is further provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0109] Those of ordinary skill in the art can understand that all or part of the processes of implementing the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories.

[0110] In summary, the multi-object trajectory generation method, system, electronic device, and storage medium provided by the present invention are directed to provided video data. Based on a specified blank video frame as a reference frame, multi-objects in each video frame are determined and position boxes are configured. The center points and coordinates of the multi-objects are determined and mapped to the video data to form continuous frame center point markings, and a multi-object movement trajectory result is formed in the video data. This can effectively solve the problem of difficult multi-object tracking and movement trajectory generation. When the multi-objects are recognized and tracked, continuous center point trajectories will be generated in the video, avoiding the situation of losing multi-objects or incorrect trajectories, and improving the accuracy of multi-object trajectory generation.

[0111] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-target object trajectory generation method, comprising: Obtain video data containing multiple target objects; Perform frame segmentation on the obtained video data, select a blank video frame as the background, determine the target objects to be tracked in the video frame after segmentation, and generate position boxes for the target objects to be tracked; Label the corresponding position boxes according to the tracking feature points of the target objects to be tracked, and calculate the center points of each position box; Obtain the center point coordinates corresponding to multiple target objects in consecutive video frames, map them to the corresponding video data according to the time sequence, and generate the moving trajectory results of multiple target objects in the video data; Determine the target objects to be tracked, including: Select a blank video frame as the background and compare it with the video frames one by one; Use the background subtraction method to identify the foreground area in the current video frame to obtain the foreground image block in the current video frame; Divide the foreground image block to determine multiple target objects to be tracked; Determine multiple target objects to be tracked, including: Obtain the foreground image block in the current video frame; Use an image recognition algorithm to identify the local feature points in the foreground image block; Obtain the difference result of the global features in the foreground image block according to the local feature points, divide the foreground image block according to the difference result, and obtain multiple target objects to be tracked; The image recognition algorithms for local feature points include: blob detection algorithm and corner detection algorithm. The blob detection algorithm includes Laplacian of Gaussian operator detection, pixel Hessian matrix, and determinant value detection; the corner detection algorithm includes Harris corner feature extraction and FAST corner feature extraction.

2. The multi-target object trajectory generation method according to claim 1, characterized in that: Generate the position box of the target object to be tracked as the contour with color and grayscale differences between the foreground image block of the target object to be tracked and the surrounding area generated according to the two types of local feature points, blobs and corners, and the generated contour forms the position box.

3. The multi-target object trajectory generation method according to claim 1, characterized in that: It also includes: Segment the foreground image blocks of each of the multiple target objects to be tracked obtained, divide the foreground image block into multiple small blocks, and calculate the histogram value and LBP feature value of each small block to form a high-dimensional vector representing the small block; Capture the target objects to be tracked according to the high-dimensional vectors of multiple small blocks.

4. A multi-target object trajectory generation system, characterized in that: The multi-target object trajectory generation system uses the multi-target object trajectory generation method described in any one of claims 1-3 to generate the moving trajectory of multiple target objects in the video data; The multi-target object trajectory generation system includes: A target object to be tracked recognition module for performing frame segmentation on the obtained video data, selecting a blank video frame as the background, and using the background subtraction method to determine the target objects to be tracked in the video frame after segmentation; A framing module for generating the contour of the target object to be tracked according to the local feature points, and the generated contour forms a position box to obtain the position box of the target object to be tracked; A center point calculation module for labeling the corresponding position boxes according to the tracking feature points of the target objects to be tracked, selecting scattered points on the contour to form a geometric figure, and calculating the center points of the geometric figures corresponding to each position box; and A trajectory generation module for mapping the center point coordinates corresponding to multiple target objects in the obtained consecutive video frames to the corresponding video data according to the time sequence to generate the moving trajectory results of multiple target objects in the video data.

5. An electronic device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor loads and executes the computer program, the steps of the method according to any one of claims 1 to 3 are implemented.

6. A storage medium storing a computer program, characterized in that, When the computer program is loaded and executed by the processor, the steps of the method according to any one of claims 1 to 3 are implemented.

Images