A UAV tracking processing method and system based on image recognition

Abstract

The present invention provides an unmanned aerial vehicle tracking processing method and system based on image recognition, which belongs to the technical field of unmanned aerial vehicles, and specifically comprises: utilizing image features of different tracking images to determine tracking target significance coefficients of different tracking images and significant target images, determining distribution data of the significant target image in the tracking image based on the acquisition time corresponding to the significant target image, and determining the tracking processing reliability of the tracking target in combination with the tracking target significance coefficients of different significant target images when the requirements are met, determining tracking interference targets of the tracking targets in different tracking images based on the image features of the tracking targets of the unmanned aerial vehicle in different tracking images, determining the tracking processing method of the unmanned aerial vehicle based on the distribution data of the tracking interference targets in different tracking images and the tracking processing reliability of the tracking targets, thereby improving the tracking processing reliability of the unmanned aerial vehicle.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicles, and in particular relates to an unmanned aerial vehicle tracking processing method and system based on image recognition. Background Art

[0002] In areas such as mines and forests, if GPS coordinates are used for tracking, only fixed targets can be tracked. In order to solve the above technical problems, the invention patent application CN202410705649.7 "A drone tracking method based on SLAM algorithm" sends the motion trajectory of the tracking target and the interface layout adjusted according to historical operations to the user terminal for display, thereby solving the problem that fixed camera tracking cannot track the object movement in real time and is difficult to adapt to complex environmental changes. However, there are the following technical problems:

[0003] When tracking a target, due to differences in the background environment and the target, there are certain differences in the interference of the drone during tracking and the reliability of the target tracking. If the tracking method cannot be determined based on the reliability of the target tracking, the reliability of the drone's tracking of the target cannot be guaranteed.

[0004] In response to the above technical problems, the present invention provides a method and system for tracking and processing unmanned aerial vehicles based on image recognition. Summary of the invention

[0005] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:

[0006] According to one aspect of the present invention, a drone tracking processing method based on image recognition is provided.

[0007] A UAV tracking processing method based on image recognition specifically includes:

[0008] S1 determines the image features of the tracking target of the UAV in different tracking images based on the analysis data of the tracking image of the UAV, and determines the tracking target significant coefficients of different tracking images and significant target images by using the image features of different tracking images;

[0009] S2: based on the acquisition time corresponding to the salient target image, determine the distribution data of the salient target image in the tracking image, and determine that the tracking processing reliability of the tracking target meets the requirements by combining the tracking target saliency coefficients of different salient target images, and then proceed to the next step;

[0010] S3 determines the tracking interference target of the tracking target in different tracking images based on the image features of the tracking target of the UAV in different tracking images;

[0011] S4 determines the distribution data of the tracking interference target in different tracking images, and determines the tracking processing method of the UAV based on the distribution data in different tracking images and the tracking processing reliability of the tracking target.

[0012] The beneficial effects of the present invention are:

[0013] Whether the tracking processing reliability of the tracking target meets the requirements is determined by the distribution data of the salient target image in the tracking image and the tracking target saliency coefficients of different salient target images, thereby achieving accurate evaluation of the tracking processing reliability of the tracking target from the perspective of the number of salient target images and the distribution data, avoiding the technical problem that the specific tracking processing method adopted by the UAV does not meet the requirements due to the reliability of the tracking processing not meeting the requirements, improving the reliability of the tracking processing, and also laying the foundation for further determining differentiated tracking processing methods for different tracking targets.

[0014] The tracking processing method of the UAV is determined based on the distribution data of different tracking images and the tracking processing reliability of the tracking target. This takes into account the differences in the reliability of the tracking processing of the UAV caused by the differences in the distribution data of the tracking interference targets in the tracking images of the UAV, and also takes into account the tracking processing reliability of the tracking target, thereby realizing the determination of differentiated tracking processing methods for different UAVs.

[0015] A further technical solution is that the tracking image of the drone is determined based on the analysis result of the collected image of the drone.

[0016] A further technical solution is that the image features include image size, color features, texture features, and shape features of the tracking target.

[0017] A further technical solution is that the method for determining the tracking target significant coefficient of the tracking image is:

[0018] Based on the image features of the tracking image, determining recognition significance coefficients corresponding to image features of different dimensions;

[0019] The tracking target saliency coefficient of the tracking image is determined according to the maximum value of the recognition saliency coefficients corresponding to the image features of different dimensions.

[0020] A further technical solution is that the identified significant coefficient is determined based on a preset significant coefficient corresponding to the image feature.

[0021] A further technical solution is that the method for determining the tracking processing method of the drone is:

[0022] Taking the tracking image with the tracking interference target as the tracking interference image, and determining the tracking interference coefficient according to the total number of the tracking interference targets in the tracking interference image;

[0023] The tracking and identification accuracy coefficient is determined based on the tracking interference coefficient and the tracking processing reliability, and the tracking and identification accuracy coefficient is used to determine the tracking processing method of the UAV.

[0024] A further technical solution is that the tracking interference coefficient is determined according to the ratio of the total number of tracking interference targets in the tracking interference image to the preset number of targets.

[0025] In a second aspect, the present invention provides a computer system comprising: a memory and a processor that are communicatively connected, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-mentioned image recognition-based drone tracking and processing method when running the computer program.

[0026] Other features and advantages will be described in the following description. The objects and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description and drawings.

[0027] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the accompanying drawings;

[0029] Figure 1 It is a flow chart of a UAV tracking processing method based on image recognition;

[0030] Figure 2 is a flow chart of a method for determining a tracking target saliency coefficient of a tracking image;

[0031] Figure 3 is a flow chart of a method for determining that the tracking processing reliability of a tracking target meets the requirements;

[0032] Figure 4 is a flow chart of a method for tracking a determination of a jamming target;

[0033] Figure 5 It is a flow chart of a method for determining a tracking processing method of a UAV;

[0034] Figure 6 It is a framework diagram of a computer system. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0036] When tracking a target, due to differences in the background environment and the target, there are certain differences in the interference situation of the drone during tracking and the reliability of the target tracking. Therefore, it is necessary to determine the tracking strategy of the target based on the interference situation of the target.

[0037] Tracking target saliency coefficient of the tracking image: Based on the image features of the tracking image, the preset saliency coefficients corresponding to the image features of different dimensions are determined, and the maximum value of the preset saliency coefficients corresponding to the image features of different dimensions is used as the tracking target saliency coefficient of the tracking image. Specifically, if the preset saliency coefficients corresponding to the image size, color feature, texture feature, and shape feature of the tracking target are 0.3, 0.5, 0.4, and 0.4, respectively, then the tracking target saliency coefficient of the tracking image is 0.5.

[0038] The tracking images with tracking target saliency coefficient greater than 0.6 are regarded as salient target images.

[0039] Tracking processing reliability of tracking targets: Divide the tracking image into multiple interval periods according to the preset interval duration, and use the interval period in which there are significant target images as the distribution period. Use the distribution data of the significant target images in different distribution periods to determine the number ratio of the significant target images in different distribution periods, and use the distribution periods with a number ratio greater than 0.5 as significant periods. Use the number ratio of significant periods to determine the tracking processing reliability, where when the number ratio of significant periods is 0.6, the tracking processing reliability is 0.6.

[0040] When the tracking process reliability is greater than 0.7, it is determined that the tracking process reliability of the tracking target meets the requirement.

[0041] Tracking interference target: The tracking interference target is determined by the similarity between the image features in the tracking image and the image features of the tracking target. Specifically, the target whose deviation between the image features in the tracking image and the image features of the tracking target is less than 0.05 is regarded as the tracking interference target.

[0042] The tracking processing method of the UAV is as follows: the tracking image with the tracking interference target is used as the tracking interference image, the tracking interference coefficient is determined by the proportion of the number of tracking interference images, the tracking recognition accuracy coefficient is determined based on the ratio of the tracking processing reliability to the tracking interference coefficient, and the tracking processing method of the UAV is determined using the tracking recognition accuracy coefficient.

[0043] UAV tracking processing method: The UAV tracking processing method is divided according to the analysis interval frequency of different images. The preset tracking processing method is to analyze and process the tracking target for each tracking image. Other tracking processing methods are to use 3 images as the analysis interval frequency and 5 images as the analysis interval frequency.

[0044] Embodiment 1 To solve the above-mentioned problem, according to one aspect of the present invention, Figure 1 As shown, a UAV tracking processing method based on image recognition is provided, which specifically includes:

[0045] S1 determines the image features of the tracking target of the UAV in different tracking images based on the analysis data of the tracking image of the UAV, and determines the tracking target significant coefficients of different tracking images and significant target images by using the image features of different tracking images;

[0046] Furthermore, the tracking image of the drone is determined based on the analysis result of the collected image of the drone.

[0047] It can be understood that the image features include the image size, color features, texture features, and shape features of the tracking target.

[0048] Specifically, Figure 2 As shown, the method for determining the tracking target significant coefficient of the tracking image is:

[0049] Based on the image features of the tracking image, determining recognition significance coefficients corresponding to image features of different dimensions;

[0050] The tracking target saliency coefficient of the tracking image is determined according to the maximum value of the recognition saliency coefficients corresponding to the image features of different dimensions.

[0051] Furthermore, the identified significance coefficient is determined based on a preset significance coefficient corresponding to the image feature.

[0052] It should be noted that when the tracking target saliency coefficient of the tracking image is greater than a preset saliency coefficient, the tracking image is determined to be a salient target image.

[0053] Optionally, the method for determining the tracking target significance coefficient of the tracking image is:

[0054] Based on the image features of the tracking image, the recognition significant coefficients corresponding to the image features of different dimensions are determined. When the recognition significant coefficients corresponding to the image features of different dimensions are all less than the preset coefficient threshold:

[0055] It is determined that the tracking image does not belong to a salient target image;

[0056] When there is an image feature that is not less than the preset coefficient threshold:

[0057] When the recognition significance coefficients of different image features are within the preset significance coefficient range:

[0058] The recognition significant coefficients of different image features are used to determine the average values ​​of the recognition significant coefficients of different image features. When the average values ​​of the recognition significant coefficients of different image features do not meet the requirements:

[0059] It is determined that the tracking image does not belong to a salient target image;

[0060] When the average value of the recognition significance coefficients of different image features meets the requirements:

[0061] The image features whose identification saliency coefficient is greater than the preset saliency coefficient are used as identification saliency features, and when the number of identification saliency features is less than the preset number of saliency features: determining that the tracking image does not belong to a saliency target image;

[0062] When the number of identified significant features is not less than the preset number of significant features:

[0063] Determining the screening significant coefficient by using different identification significant coefficients for identifying significant features, and when the screening significant coefficient does not meet the requirement, determining that the tracking image does not belong to the significant target image;

[0064] When the screening significant coefficient meets the requirements or there are image features that are not within the preset significant coefficient range:

[0065] The recognition saliency coefficients of different image features are obtained, and the tracking target saliency coefficient of the tracking image is determined in combination with the screening saliency coefficient.

[0066] S2: based on the acquisition time corresponding to the salient target image, determine the distribution data of the salient target image in the tracking image, and determine that the tracking processing reliability of the tracking target meets the requirements by combining the tracking target saliency coefficients of different salient target images, and then proceed to the next step;

[0067] Further, the distribution data of the salient target image in the tracking image includes a distribution period in the tracking image and distribution data of the corresponding distribution period.

[0068] Specifically, Figure 3 As shown, determining that the tracking processing reliability of the tracking target meets the requirements specifically includes:

[0069] Dividing the tracking image into a plurality of interval time periods according to a preset interval duration, and taking the interval time period in which a significant target image exists as a distribution time period;

[0070] Determine the quantity ratio of the significant target images in different distribution time periods by using the distribution data of the significant target images in different distribution time periods, and determine the significant time period in the distribution time period according to the quantity ratio;

[0071] The reliability of the tracking process of the tracking target is determined according to the number of significant time periods, and whether the reliability of the tracking process of the tracking target meets the requirement is determined in combination with a preset reliability threshold.

[0072] Optionally, when the tracking processing reliability of the tracking target is less than a preset reliability threshold, it is determined that the tracking processing reliability of the tracking target does not meet the requirement.

[0073] Further, when the tracking processing reliability of the tracking target does not meet the requirement, the tracking processing of the tracking target is performed according to a preset tracking processing method.

[0074] It should be noted that determining that the tracking processing reliability of the tracking target meets the requirements specifically includes:

[0075] Determine the quantity ratio of the salient target image in the tracking image by using the distribution data of the salient target image, and determine the image scale factor of the salient target image by using the quantity ratio;

[0076] Determine the average value of the tracking target saliency coefficients using the tracking target saliency coefficients of different saliency target images, and use it as the saliency coefficient mean;

[0077] The reliability of the tracking process of the tracking target is determined based on the product of the significant coefficient mean and the image scale factor, and combined with a preset reliability threshold, it is determined whether the reliability of the tracking process of the tracking target meets the requirement.

[0078] Optionally, determining whether the tracking processing reliability of the tracking target meets the requirement specifically includes:

[0079] S21 divides the tracking image into a plurality of interval time periods according to a preset interval duration, and uses the interval time period in which a significant target image exists as a distribution time period;

[0080] Optionally, the above step S21 includes the following contents:

[0081] S211 divides the tracking image into a plurality of interval periods according to a preset interval duration, and uses the interval period in which a significant target image exists as a distribution period. When the number of distribution periods is less than the preset number of distribution periods, it is determined that the tracking processing reliability of the tracking target does not meet the requirement. When the number of distribution periods is not less than the preset number of distribution periods, the process proceeds to step S212.

[0082] S212: obtaining the number of salient target images in the tracking image, and when the number of salient target images is greater than the preset number of targets, proceeding to step S22; when the number of salient target images is not greater than the preset number of targets, proceeding to step S213;

[0083] S213: When the number of the distribution time periods is within the preset time period number range, it is determined that the tracking processing reliability of the tracking target does not meet the requirement; when the number of the distribution time periods is not within the preset time period number range, the process goes to step S22.

[0084] S22 determines the quantity ratio of the salient target images in different distribution time periods by using the distribution data of the salient target images in different distribution time periods, and determines the time period significance coefficient of the distribution time period according to the quantity ratio and the tracking target significance coefficient of the salient target images in the distribution time period;

[0085] Optionally, the above step S22 includes the following contents:

[0086] S221 determines a basic saliency coefficient based on the proportion of the number of salient target images in the tracking image. When the basic saliency coefficient is less than a preset saliency coefficient threshold, the process proceeds to step S222. When the basic saliency coefficient is not less than the preset saliency coefficient threshold, the process proceeds to step S223.

[0087] S222 determines the quantity ratio of the salient target images in different distribution time periods by using the distribution data of the salient target images in different distribution time periods. When there is a distribution time period in which the quantity ratio meets the requirement, the process proceeds to step S223. When there is no distribution time period in which the quantity ratio meets the requirement, it is determined that the tracking processing reliability of the tracking target does not meet the requirement.

[0088] S223 determines the time period significance coefficient of the distribution period according to the number proportion and the tracking target significance coefficient of the significant target image in the distribution period, and when there is no distribution period with a time period significance coefficient greater than a preset significance threshold, proceeds to step S224; when there is a distribution period with a time period significance coefficient greater than a preset significance threshold, proceeds to step S23;

[0089] S224 When the average value of the period significance coefficients of different distribution time periods is within the preset significance coefficient range, it is determined that the tracking processing reliability of the tracking target does not meet the requirements; when the average value of the period significance coefficients of different distribution time periods is not within the preset significance coefficient range, proceed to step S23.

[0090] S23 determines the tracking processing reliability of the tracking target according to the time period significance coefficients of different distribution time periods, and determines whether the tracking processing reliability of the tracking target meets the requirements in combination with a preset reliability threshold.

[0091] S3 determines the tracking interference target of the tracking target in different tracking images based on the image features of the tracking target of the UAV in different tracking images;

[0092] Specifically, Figure 4 As shown, the method for determining the tracking interference target is:

[0093] Determine image features of other objects in the tracking image based on image features of the tracking target of the drone in the tracking image;

[0094] Determining a similarity coefficient between the image features of the other object and the image features of the tracking target according to similarities between the image features of the other object and the image features of the tracking target in the tracking image;

[0095] The image feature similarity coefficient is used to determine whether the other object is a tracking interference target.

[0096] Furthermore, when the image feature similarity coefficient of the other object meets the requirement, the other object is determined to be a tracking interference target.

[0097] S4 determines the distribution data of the tracking interference target in different tracking images, and determines the tracking processing method of the UAV based on the distribution data in different tracking images and the tracking processing reliability of the tracking target.

[0098] It should be noted that if Figure 5 As shown, the method for determining the tracking processing method of the drone is:

[0099] The tracking image containing the tracking interference target is regarded as the tracking interference image, and the number ratio of the tracking interference images is determined based on the distribution data of the tracking interference images;

[0100] Determine the image interference coefficient according to the proportion of the number of tracking interference images;

[0101] The tracking and identification accuracy coefficient is determined based on the image interference coefficient and the tracking processing reliability, and the tracking and identification accuracy coefficient is used to determine the tracking processing method of the UAV.

[0102] Furthermore, the tracking and identification accuracy coefficient is determined according to the ratio of the tracking processing reliability to the image interference coefficient.

[0103] Optionally, the tracking and identification accuracy coefficient is used to determine the tracking processing method of the drone, specifically including:

[0104] The preset tracking processing method corresponding to the tracking identification accuracy coefficient is determined by the tracking identification accuracy coefficient, and the tracking processing method of the UAV is determined by using the corresponding preset tracking processing method.

[0105] Optionally, the method for determining the tracking processing method of the drone is:

[0106] Taking the tracking image with the tracking interference target as the tracking interference image, and determining the tracking interference coefficient according to the total number of the tracking interference targets in the tracking interference image;

[0107] The tracking and identification accuracy coefficient is determined based on the tracking interference coefficient and the tracking processing reliability, and the tracking and identification accuracy coefficient is used to determine the tracking processing method of the UAV.

[0108] Furthermore, the tracking interference coefficient is determined according to a ratio of a total number of tracking interference targets in the tracking interference image to a preset number of targets.

[0109] Optionally, the method for determining the tracking processing method of the drone is:

[0110] S41: taking the tracking image with the tracking interference target as the tracking interference image, dividing the tracking image into a plurality of interval time periods according to a preset interval time, and taking the interval time period with the tracking interference image as the interference distribution time period;

[0111] Optionally, the above step S41 includes the following contents:

[0112] S411: taking the tracking images with the tracking interference targets as the tracking interference images. When the number of the tracking interference images does not meet the requirement, the tracking processing of the tracking targets is performed using the preset tracking processing method. When the number of the tracking interference images meets the requirement, the process proceeds to step S412.

[0113] S412 determines the screening interference image based on the number of tracking interference targets in the tracking interference image. When the number of the screening interference images does not meet the requirement, the tracking target is tracked using a preset tracking processing method. When the number of the tracking interference images meets the requirement, the process proceeds to step S413.

[0114] S413 determines the basic interference coefficient according to the number of tracking interference targets in different tracking interference images. When the basic interference coefficient does not meet the requirement, the tracking process of the tracking target is performed using a preset tracking process method. When the basic interference coefficient meets the requirement, the process proceeds to step S414.

[0115] S414 divides the tracking image into multiple interval periods according to the preset interval duration, takes the interval period where the tracking interference image exists as the interference distribution period, and proceeds to step S42.

[0116] S42 determines interference coefficients for different interference distribution periods according to the proportion of the number of tracking interference images in different interference distribution periods and the number of tracking interference targets in different tracking interference images;

[0117] Optionally, the above step S42 includes the following contents:

[0118] S421: obtaining the number of interference distribution periods. When the number of interference distribution periods does not meet the requirement, the tracking process of the tracking target is performed using a preset tracking process method. When the number of interference distribution periods meets the requirement, the process proceeds to step S422.

[0119] S422 determines the abnormal interference distribution period according to the proportion of the number of tracking interference images in the interference distribution period. When the number of abnormal interference distribution periods does not meet the requirement, the tracking process of the tracking target is performed using a preset tracking process method. When the number of abnormal interference distribution periods meets the requirement, the process proceeds to step S423.

[0120] S423 determines the interference coefficients of different interference distribution time periods based on the proportion of the number of tracking interference images in different interference distribution time periods and the number of tracking interference targets in different tracking interference images. When there are a preset number of interference distribution time periods whose interference coefficients do not meet the requirements, the preset tracking processing method is used to track the tracking targets. When the number of interference distribution time periods whose interference coefficients do not meet the requirements is within the preset time period number range, proceed to step S43.

[0121] S43 determines the image interference coefficient based on the interference coefficients of different interference distribution periods, determines the tracking and identification accuracy coefficient through the image interference coefficient and the tracking processing reliability, and determines the tracking processing method of the UAV using the tracking and identification accuracy coefficient.

[0122] Embodiment 2 The second aspect, as Figure 6As shown, the present invention provides a computer system, comprising: a memory and a processor that are communicatively connected, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-mentioned image recognition-based drone tracking processing method when running the computer program.

[0123] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0124] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0125] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. A drone tracking processing method based on image recognition, characterized in that: Specifically include: Determine the image features of the tracking target of the UAV in different tracking images based on the analysis data of the tracking image of the UAV, and determine the tracking target significance coefficients of different tracking images and the significant target images using the image features of different tracking images; Based on the acquisition time corresponding to the salient target image, the distribution data of the salient target image in the tracking image is determined, and when the tracking target saliency coefficients of different salient target images are combined to determine that the tracking processing reliability of the tracking target meets the requirements, the process proceeds to the next step; Determining tracking interference targets of the tracking targets in different tracking images based on image features of the tracking targets of the UAV in different tracking images; Determine distribution data of the tracking interference target in different tracking images, and determine the tracking processing method of the UAV based on the distribution data in different tracking images and the tracking processing reliability of the tracking target; The method for determining the tracking target significant coefficient of the tracking image is: Based on the image features of the tracking image, determining recognition significance coefficients corresponding to image features of different dimensions; The tracking target saliency coefficient of the tracking image is determined according to the maximum value of the recognition saliency coefficients corresponding to the image features of different dimensions.

2. The method for tracking and processing unmanned aerial vehicles based on image recognition according to claim 1, characterized in that: The tracking image of the UAV is determined according to the analysis result of the collected image of the UAV.

3. The method for tracking and processing unmanned aerial vehicles based on image recognition according to claim 1, characterized in that: The image features include image size, color features, texture features, and shape features of the tracking target.

4. The method for tracking and processing unmanned aerial vehicles based on image recognition according to claim 1, characterized in that: The identified saliency coefficient is determined based on a preset saliency coefficient corresponding to the image feature.

5. The method for tracking and processing unmanned aerial vehicles based on image recognition according to claim 1, characterized in that: When the tracking target saliency coefficient of the tracking image is greater than a preset saliency coefficient, the tracking image is determined to be a salient target image.

6. The method for tracking and processing unmanned aerial vehicles based on image recognition according to claim 1, characterized in that: The distribution data of the salient target image in the tracking image includes a distribution period in the tracking image and distribution data of the corresponding distribution period.

7. The method for tracking and processing unmanned aerial vehicles based on image recognition according to claim 1, characterized in that: The method for determining the tracking interference target is: Determine image features of other objects in the tracking image based on image features of the tracking target of the drone in the tracking image; Determining a similarity coefficient between the image features of the other object and the image features of the tracking target according to similarities between the image features of the other object and the image features of the tracking target in the tracking image; The image feature similarity coefficient is used to determine whether the other object is a tracking interference target.

8. The method for tracking and processing unmanned aerial vehicles based on image recognition according to claim 7, characterized in that: When the image feature similarity coefficient of the other object meets the requirement, the other object is determined to be a tracking interference target.

9. A computer system comprising: A memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes a drone tracking and processing method based on image recognition as described in any one of claims 1-8 when running the computer program.

Images