Image sequence analysis method for dynamic target trajectory prediction of seeker

Abstract

This invention provides an image sequence analysis method for predicting the trajectory of a dynamic target using a seeker, belonging to the field of image processing. The method includes: obtaining all target regions within each target image based on the matching degree between each region in each image captured by the seeker and images of the target at multiple angles when it is not in flight, and the distribution of other regions surrounding each region in each target image captured by the seeker; combining the real-time velocity of the seeker when capturing each target image with the previous target image, and the real-time velocity of the target when acquiring the first target image, to complete the trajectory prediction of the target. This invention aims to solve the problem that current vision-based methods for predicting the trajectory of targets tracked by seekers suffer from relatively low accuracy due to interference from complex environmental factors such as clouds.

Description

Technical Field

[0001] This invention belongs to the field of image processing, specifically relating to an image sequence analysis method for predicting the dynamic target trajectory of a seeker. Background Technology

[0002] Given the urgent need for modern autonomous guidance, researching dynamic target trajectory prediction methods based on image sequence analysis is of paramount importance. Predicting the dynamic trajectory of the seeker not only significantly improves the accuracy and autonomous operation capability of precision guidance systems in highly contested environments, but also strongly promotes interdisciplinary innovation in computer vision, artificial intelligence, and guidance control.

[0003] However, in traditional vision methods, when the seeker is tracking a target, the accuracy of predicting the trajectory of the dynamic target being tracked is relatively low due to interference from complex factors in the environment such as clouds and smoke, as well as strong interference points such as decoys. This results in a decrease in trajectory tracking accuracy and consequently, a relatively low accuracy in predicting the trajectory of the tracked target. Summary of the Invention

[0004] To address the issue of relatively low accuracy in predicting the trajectory of a target tracked by a seeker using traditional visual methods, this invention proposes an image sequence analysis method for predicting the dynamic target trajectory of a seeker.

[0005] To achieve the above objectives, the present invention provides the following technical solution: acquiring images of the target being tracked from multiple angles when it is not in flight, and a sequence of target image sequences captured by the seeker; obtaining the real-time speed and position of the seeker when acquiring each target image, and the real-time speed of the target when acquiring the first target image; using an edge detection algorithm to obtain the regions within each target image; and based on the matching degree between each region within each target image captured by the seeker and the target being tracked from multiple angles when it is not in flight, and the difference in the matching degree between different regions within the image and the target being tracked from multiple angles when it is not in flight, determining the reliability of each region within each target image captured by the seeker as a target region. Then, the suspected target region within each tracking target image is obtained; based on the distribution of suspected target regions around each suspected target region in each tracking target image captured by the seeker, and the difference in the confidence of each suspected target region and its surrounding suspected target regions as tracking target regions, all target regions within each tracking target image are obtained; based on the area difference of all target regions in each tracking target image captured by the seeker and the previous tracking target image, the real-time speed of the seeker when capturing each tracking target image and the previous tracking target image, and the real-time speed of the tracking target when acquiring the first tracking target image, the real-time speed of the tracking target when the seeker captures each tracking target image is obtained, thus completing the trajectory prediction of the tracking target.

[0006] Furthermore, the specific steps for obtaining the reliability of each region within each tracking target image captured by the seeker as the tracking target region are as follows: Obtain the first image captured by the seeker... The first image within the tracking target The area and the tracked target were in the non-flying state. The maximum matching degree among all images at the nth angle is denoted as the nth image captured by the seeker. The first image within the tracking target The degree of matching between each region and the tracking target; among which, The preset number of angles; the number of images captured by the seeker head. The first image within the tracking target The area and the tracked target were in the non-flying state. The image of the target when it is not in flight, corresponding to the maximum matching degree among all images at all angles, is denoted as the image of the target captured by the seeker. The first image within the tracking target The image corresponding to each region; the first image captured by the seeker head. The first image within the tracking target The area and the first image captured by the guide head The first image within the tracking target The matching degree of the image corresponding to the region is denoted as the first image captured by the seeker. The first image within the tracking target The region is in the first Matching degree based on a region; if the seeker captures the first... The first image within the tracking target The degree of matching between each area and the tracked target, and the first image captured by the seeker. The first image within the tracking target The region is in the first If the absolute value of the difference in matching degree based on each region is less than or equal to 0.1, then the first image captured by the seeker will be... The first image within the tracking target The area is designated as the first area captured by the guidance head. The first image within the tracking target Similar areas in each region; based on the images captured by the seeker head... The first image within the tracking target Similar areas in each region, and the first image captured by the seeker. The first image within the tracking target The degree of matching between each region and the tracked target is used to obtain the first image captured by the seeker. The first image within the tracking target Each region is used to track the credibility of the target region.

[0007] Furthermore, the first image captured by the seeker head... The first image within the tracking target The specific formula for calculating the credibility of the target area for each region is as follows:

[0008] ;

[0009] In the formula, Indicates the first image captured by the seeker. The first image within the tracking target Each region is used to track the credibility of the target region. Indicates the first image captured by the seeker. The first image within the tracking target The degree of matching between each region and the tracking target. Indicates the first image captured by the seeker. The number of regions within the target image being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The number of similar regions in each region To prevent hyperparameters with a denominator of 0, This represents the sigmoid function.

[0010] Furthermore, the specific steps for obtaining the suspected target region within each tracking target image are as follows: Preset a confidence threshold. If the seeker head captures the first... The first image within the tracking target The credibility of each region as the target region is greater than [a certain value]. Then the first image captured by the seeker head will be... The first image within the tracking target Each area is designated as a suspected target area.

[0011] Furthermore, the specific steps for obtaining all target regions within each tracking target image are as follows: Acquire the first image captured by the seeker... The first image within the tracking target The suspected target area and the first The centroid of a suspected target area; using the convex hull method, in the first image captured by the seeker... Within the target tracking image, generate a second image that simultaneously contains images captured by the seeker head. The first image within the tracking target The suspected target area and the first The area suspected to be the target area was marked as the first area captured by the seeker head. The first image within the tracking target The suspected target area and the first A connected area of ​​a suspected target region; based on the image captured by the seeker head... The first image within the tracking target The suspected target area and the first The connected area of ​​the suspected target area does not belong to the first... The suspected target area and the first The area of ​​other areas in the suspected target area, the first image captured by the seeker head The first image within the tracking target The suspected target area and the first The area of ​​the connected region of the suspected target area, the first image captured by the seeker. The first image within the tracking target The suspected target area and the first The Euclidean distance of the centroid of the suspected target area, and the first image captured by the seeker. The first image within the tracking target The suspected target area and the first The difference in the credibility of the suspected target area is due to the discrepancy in the reliability of the target area, and the result is the first image captured by the seeker. The first image within the tracking target The suspected target area and the first Each suspected target area contributes to the credibility of the tracked target; based on the images captured by the seeker... The first image within the tracking target The suspected target area and the first Each suspected target area contributes to the credibility of the tracked target, determining the reliability of the image captured by the seeker. The first image within the tracking target The suspected target area and the first Is the suspected target area the target area?

[0012] Furthermore, the first image captured by the seeker head... The first image within the tracking target The suspected target area and the first The specific formula for calculating the confidence level that all suspected target areas are tracking targets is as follows:

[0013] ;

[0014] In the formula, Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first Each suspected target area contributes to the credibility of the tracked target. Indicates the first image captured by the seeker. The first image within the tracking target The centroid of the suspected target area and the first The Euclidean distance of the centroid of a suspected target region. Indicates the first image captured by the seeker. The first image within the tracking target The number of suspected target areas is used to assess the credibility of the target area being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The number of suspected target areas is used to assess the credibility of the target area being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first The area of ​​the connected regions of a suspected target area. Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first In the connected area of ​​the suspected target area, except for the first The suspected target area and the first The area of ​​other areas outside the suspected target area, To prevent hyperparameters with a denominator of 0, It is an exponential function with the natural constant as its base. This represents the sigmoid function. This represents the absolute value function.

[0015] Furthermore, the judgment of the first image captured by the seeker head... The first image within the tracking target The suspected target area and the first The specific steps for determining whether a suspected target area is indeed a target area are as follows: Preset a confidence threshold. If the seeker head captures the first... The first image within the tracking target The suspected target area and the first All suspected target areas have a confidence level greater than the confidence threshold for tracking targets. Then the first image captured by the seeker head will be... The first image within the tracking target The suspected target area and the first All suspected target areas are recorded as target areas.

[0016] Further, the specific steps for obtaining the real-time speed of the tracking target when the seeker captures each tracking target image are as follows: Based on the real-time speed of the tracking target when the seeker captures the first tracking target image, the capture time of the seeker capturing the first and second tracking target images, the real-time speed of the seeker when capturing the first and second tracking target images, and the sum of the areas of all target regions in the first and second tracking target images, the real-time speed of the tracking target when the seeker captures the second tracking target image is obtained; based on the method for obtaining the real-time speed of the tracking target when the seeker captures the second tracking target image, the real-time speed of the tracking target when the seeker captures the third tracking target image is obtained, and the above operations are repeated continuously to obtain the real-time speed of the tracking target captured by the seeker in the first tracking target image. The real-time speed at which the target is tracked when tracking an image of the target.

[0017] Furthermore, the specific formula for calculating the real-time speed of the tracking target when the seeker captures the second image of the tracking target is as follows:

[0018] ;

[0019] In the formula, Indicates the seeker head taking the first shot The real-time speed at which the target is tracked when tracking the target image. Indicates the seeker head taking the first shot The real-time speed at which the target is tracked when tracking the target image. Indicates the seeker head taking the first shot The time it took to capture the image of the target being tracked. Indicates the seeker head taking the first shot The real-time speed of the seeker when tracking a target image. Indicates the seeker head taking the first shot The real-time speed of the seeker when tracking a target image. Indicates the seeker head taking the first shot The time it took to capture the image of the target being tracked. Indicates the first image captured by the seeker. The sum of the areas of all target regions within the image of the target being tracked. Indicates the first image captured by the seeker. The sum of the areas of all target regions within the image of the target being tracked. This represents the sigmoid function. It is an exponential function with the natural constant as its base.

[0020] Furthermore, the specific steps for completing the trajectory prediction of the tracking target are as follows: The seeker head captures the first... When tracking a target image, the real-time velocity of the target minus the velocity captured by the seeker in the first image is used to track the target. The difference between the real-time velocity of the target and the velocity of the target when the seeker captures the first image is denoted as the value of the difference between the real-time velocity of the target and the velocity of the target when the seeker captures the first image. The acceleration of the target is measured when the first image of the target is captured; a three-dimensional coordinate system is constructed with the position of the seeker when the first image of the target is captured as the origin, and the position and velocity of the seeker are obtained when each image of the target is captured; based on the image captured by the seeker... The position of the seeker head in the three-dimensional coordinate system when tracking target images, based on the first image captured by the seeker head. Using the lower left corner of the target image as the origin, a two-dimensional coordinate system is constructed, and the image captured by the seeker within this two-dimensional coordinate system is obtained. The coordinates of the centroid of a region formed by all target areas within the target image are used in conjunction with the seeker's image capture. When tracking target images, the acceleration and velocity of the target are monitored. Using existing dynamic models, the acceleration and velocity of the target are predicted for the first image captured by the seeker. Tracking the target's trajectory when tracking the target image.

[0021] The image sequence analysis method for dynamic target trajectory prediction of a seeker provided by this invention has the following beneficial effects: When obtaining the tracking target from images captured by the seeker, this invention first divides the image into multiple regions based on the characteristic that the tracking target in the image captured by the seeker has a large difference from the background region. Then, based on the characteristics that the region segmented by the tracking target has a high matching degree with the image of the tracking target in a static state, and that the number of regions segmented by the tracking target in the image is relatively small, a suspected target region is obtained from the region segmented by each image. Then, based on the distribution of other suspected target regions around each suspected target region in each image, the tracking target in each image is obtained. The invention lays the groundwork for subsequent trajectory prediction of the tracked target. Then, based on the real-time speed of the seeker when taking each image and the size changes of the tracked target in the captured image, and according to the principle of near objects appearing larger and farther objects smaller, the real-time speed of the tracked target is obtained. A three-dimensional coordinate system is then constructed. Based on the seeker's position when taking each image, the real-time speed of the tracked target within the image, and the real-time speed of the tracked target, trajectory prediction is completed. This solves the problem that in traditional vision methods, when the seeker tracks a target, the complexity of the environment and strong interference points such as decoys cause abnormal trajectory prediction of the dynamic target being tracked, resulting in decreased trajectory tracking accuracy. Attached Figure Description

[0022] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of an image sequence analysis method for dynamic target trajectory prediction by a seeker, according to an embodiment of the present invention. Detailed Implementation

[0024] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0025] Example 1: This invention provides an image sequence analysis method for predicting the dynamic target trajectory of a seeker, specifically as follows: Figure 1As shown, the process includes: Step S001: acquiring images of the target at multiple angles when it is not in flight and a sequence of images of the target captured by the seeker, and obtaining the real-time speed and real-time position of the seeker when acquiring each image of the target and the real-time speed of the target when acquiring the first image of the target.

[0026] Specifically, the data collection interval is set to... Every second, the camera inside the seeker head captures multiple images of the tracked target in real time. Based on the speed and position sensors within the seeker head, the sensors simultaneously acquire the real-time speed and position of the seeker head while capturing images of the tracked target. This yields the real-time speed and position of the seeker head at the time of acquiring each image of the tracked target. The preset acquisition interval in this embodiment... This example is used for illustration; other values ​​may be used in other implementations, and this embodiment is not limited to any particular value. Furthermore, in this embodiment, the images obtained by the seeker, as well as the images of the tracked target when it is not in flight, are both black and white images.

[0027] Furthermore, acquiring information about the tracking target when it is not in flight. Images from multiple angles are obtained, showing the tracked target at multiple angles when it is not in flight. The preset number of angles in this embodiment... This example is used to illustrate the concept; other values ​​can be set in other implementations.

[0028] Furthermore, using existing technology, the real-time tracking speed of the seeker when taking the first image of the target is obtained. That is, the real-time tracking speed when acquiring the first image of the target is obtained.

[0029] Thus, the sequence of images of the target captured by the seeker is obtained, the real-time speed and position of the seeker are obtained for each target image, and the real-time speed of the target is obtained when the first target image is captured.

[0030] Step S002: Use an edge detection algorithm to obtain the region within each tracking target image; based on the matching degree between each region within each tracking target image captured by the seeker and the images of the tracking target at multiple angles when it is not in flight, and the difference in the matching degree between different regions within the image and the images of the tracking target at multiple angles when it is not in flight, the credibility of each region within each tracking target image captured by the seeker as a tracking target region is obtained, and thus the suspected target region within each tracking target image is obtained.

[0031] It should be noted that the prediction of dynamic targets by the seeker is often based on the real-time target trajectory to predict the target trajectory at subsequent time points. Considering the complex environmental factors such as clouds and smoke during flight, the seeker's analysis of the tracked target's position may be inaccurate, thus affecting the accuracy of subsequent target trajectory analysis. Therefore, when predicting the trajectory of a dynamic target, the first step is to obtain the dynamic target from the images captured by the seeker. Then, the dynamic target in the images is analyzed to complete the trajectory prediction.

[0032] It should be further noted that the images captured by the seeker are variable; that is, the target state in the captured images can vary, and there may be deviations from the target image set in the system. This makes it impossible to directly use feature comparison to obtain the position of the tracking target in the image acquired by the seeker. Therefore, this invention, based on the characteristic that there is a certain difference between the tracking target in the image and the background, uses an edge detection algorithm to divide the collected single images into regions.

[0033] It should be further noted that after dividing each image into regions, the tracking target may not exist in a single region due to interference. Therefore, the specific appearance of each region in each image is combined to identify and analyze the suspected target region in a single image, in order to find the region where the tracking target is located in a single image.

[0034] It should be further noted that, since the region containing the tracking target in a single image has a high matching degree with the pre-acquired tracking target, this invention compares and matches the region with the multi-angle photos of the tracking target obtained from the acquisition, and selects the region with the highest matching degree as the tracking target matching degree. Then, based on the tracking target matching degree of each region, the reliability of each region as the target region is calculated.

[0035] It should be further noted that the number of target regions in a single image is relatively small compared to the number of background regions. Furthermore, the tracking target matching degree of background regions differs significantly from that of target regions in a single image. Therefore, based on the tracking target matching degree of each region in a single image, and the difference between the tracking target matching degree of each region and the tracking target matching degree of other regions, the credibility of each region as a target region is determined, thus identifying potential target regions.

[0036] Specifically, using edge detection algorithms, the first image captured by the seeker is... Edge detection is performed on the target image captured by the seeker to obtain the first image. Zhang tracks multiple edges within the target image. Based on the first image captured by the seeker... Zhang traces the edges within the target image to obtain the first image captured by the seeker. Zhang tracks all regions within the target image. In this embodiment, the region within the image refers to the area enclosed by the edges in the image, and different regions in the image do not contain overlapping parts.

[0037] Furthermore, the first image captured by the seeker head... The first image within the tracking target The area and the tracked target were in the non-flying state. The maximum matching degree among all images at the nth angle is denoted as the nth image captured by the seeker. The first image within the tracking target The degree of matching between each region and the tracking target.

[0038] Furthermore, the first image captured by the seeker head... The first image within the tracking target The area and the tracked target were in the non-flying state. The image of the target when it is not in flight, corresponding to the maximum matching degree among all images at all angles, is denoted as the image of the target captured by the seeker. The first image within the tracking target The image corresponding to each region.

[0039] Furthermore, the first image captured by the seeker head... The first image within the tracking target The area and the first image captured by the guide head The first image within the tracking target The matching degree of the image corresponding to the region is denoted as the first image captured by the seeker. The first image within the tracking target The region is in the first Matching degree based on regions. Obtaining the matching degree between a region in one image and another image is a well-known existing technique, and will not be elaborated upon in this embodiment.

[0040] Furthermore, if the seeker head captures the first... The first image within the tracking target The degree of matching between each area and the tracked target, and the first image captured by the seeker. The first image within the tracking target The region is in the first If the absolute value of the difference in matching degree based on each region is less than or equal to 0.1, then the first image captured by the seeker will be... The first image within the tracking target The area is designated as the first area captured by the guidance head. The first image within the tracking target Similar regions within each region. Among them... In this embodiment, the threshold value used in this section is 0.1; in other embodiments, it can be set to other values.

[0041] Furthermore, the first image captured by the seeker head... The first image within the tracking target The specific formula for calculating the credibility of the target area for each region is as follows:

[0042] ;

[0043] In the formula, Indicates the first image captured by the seeker. The first image within the tracking target Each region is used to track the credibility of the target region. Indicates the first image captured by the seeker. The first image within the tracking target The degree of matching between each region and the tracking target. Indicates the first image captured by the seeker. The number of regions within the target image being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The number of similar regions in each region To prevent hyperparameters with a denominator of 0, this embodiment sets... , This represents the sigmoid function, which is used for normalization in this embodiment.

[0044] It should be noted that, The larger the value, the higher the number of images captured by the seeker. The first image within the tracking target The greater the match between a region and the tracked target, the more it indicates that the seeker's image captures the first... The first image within the tracking target The more reliable the region is as a target area, the stronger its reliability. It should be noted that, due to the image captured by the seeker... The number of background areas within the target image is relatively large, making it difficult for the seeker to capture the first... The first image within the tracking target When each region is the target region The value is small. The value is relatively large, that is The larger the value, the higher the number of images captured by the seeker. The first image within the tracking target The more reliable a region is for tracking the target region, the stronger its credibility.

[0045] Furthermore, a preset credibility threshold is established. If the seeker head captures the first... The first image within the tracking target The credibility of each region as the target region is greater than [a certain value]. Then the first image captured by the seeker head will be... The first image within the tracking target Each region is designated as a suspected target region. The confidence threshold preset in this embodiment... This example is used to illustrate the concept; other values ​​can be set in other implementations.

[0046] At this point, the suspected target area within each image of the tracked target captured by the seeker is obtained.

[0047] Step S003: Based on the distribution of suspected target regions around each suspected target region in each tracking target image captured by the seeker, and the difference in the credibility of each suspected target region and its surrounding suspected target regions as tracking target regions, obtain all target regions in each tracking target image.

[0048] It should be noted that after obtaining the suspected target region in each image, the tracking target in the image is very likely to be divided into multiple regions during image segmentation. Therefore, this invention obtains the final target region from multiple suspected target regions.

[0049] It should be further noted that if two suspected target regions in an image are both part of the tracked target, then the credibility of these two regions as target regions is relatively similar, because both regions are part of the target region. Therefore, based on the difference in credibility between each suspected target region and other suspected target regions in the same image, it can be determined whether these two suspected target regions are simultaneously part of the tracked target.

[0050] It should be further explained that since the tracking target is a whole, when performing region segmentation on the image, if the tracking target is divided into multiple regions, then each region to which the tracking target is divided must contain other regions to which the tracking target is divided. Therefore, the final target region in the image is obtained based on the distribution of other suspected target regions around each suspected target region.

[0051] Specifically, acquiring the first image captured by the seeker head. The first image within the tracking target The suspected target area and the first The centroid of a suspected target region. Obtaining the centroid of a region in an image is a well-known technique and will not be described in detail in this embodiment.

[0052] Furthermore, using the convex hull method, in the first image captured by the seeker... Within the target tracking image, generate a second image that simultaneously contains images captured by the seeker head. The first image within the tracking target The suspected target area and the first The area suspected to be the target area was marked as the first area captured by the seeker head. The first image within the tracking target The suspected target area and the first A connected region of suspected target area. The method of generating a region using the convex hull method is a well-known existing technique and will not be elaborated upon in this embodiment.

[0053] Furthermore, the first image captured by the seeker head... The first image within the tracking target The suspected target area and the first The specific formula for calculating the confidence level that all suspected target areas are tracking targets is as follows:

[0054] ;

[0055] In the formula, Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first Each suspected target area contributes to the credibility of the tracked target. Indicates the first image captured by the seeker. The first image within the tracking target The centroid of the suspected target area and the first The Euclidean distance of the centroid of a suspected target region. Indicates the first image captured by the seeker. The first image within the tracking target The number of suspected target areas is used to assess the credibility of the target area being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The number of suspected target areas is used to assess the credibility of the target area being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first The area of ​​the connected regions of a suspected target area. Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first In the connected area of ​​the suspected target area, except for the first The suspected target area and the first The area of ​​other areas outside the suspected target area, To prevent hyperparameters with a denominator of 0, this embodiment sets... This example is used for illustration; other values ​​can be set in other implementations. As an exponential function with the natural constant as its base, this embodiment uses it to represent an inverse proportional relationship; This represents the sigmoid function, which is used in this embodiment for normalization. This represents the absolute value function.

[0056] It should be noted that, The smaller the value, the more likely it is that the seeker has captured the first image. The first image within the tracking target The suspected target area and the first The two suspected target areas are quite close, further illustrating the significance of the seeker's image. The first image within the tracking target The suspected target area and the first The distribution of suspected target areas is more consistent with the characteristics of each region divided by the tracked target in the image of the tracked target captured by the seeker, which will contain multiple other regions divided by the tracked target around each region divided by the tracked target. The smaller the value, the more likely it is that the seeker has captured the first image. The first image within the tracking target The suspected target area and the first The two suspected target areas are quite close, further illustrating the significance of the seeker's image. The first image within the tracking target The suspected target area and the first There is a high probability that all suspected target areas are the targets being tracked. The smaller the value, the more likely it is that the seeker has captured the first image. The first image within the tracking target The suspected target area and the first The more a suspected target area matches the characteristic that if two areas both belong to the target area being tracked, then the more similar the credibility of the two areas being the target area being tracked.

[0057] Furthermore, a preset credibility threshold is set. If the seeker head captures the first... The first image within the tracking target The suspected target area and the first All suspected target areas have a confidence level greater than the confidence threshold for tracking targets. Then the first image captured by the seeker head will be... The first image within the tracking target The suspected target area and the first All suspected target regions are recorded as target regions. In this embodiment, a preset confidence threshold is used. This example will be used to illustrate the concept; other implementations may use different values. Specifically, if the seeker captures the first... If there is only one suspected target area in the image of the tracked target, then that suspected target area is recorded as the first area captured by the seeker. Zhang tracks the target region within the target image.

[0058] At this point, all target areas within each image of the tracking target captured by the seeker are obtained.

[0059] Step S004: Based on the area difference of all target regions in each image captured by the seeker and the previous image of the target being tracked, the real-time speed of the seeker when capturing each image of the target being tracked and the previous image of the target being tracked, and the real-time speed of the target being tracked when acquiring the first image of the target being tracked, the real-time speed of the target being tracked when the seeker captures each image of the target being tracked is obtained, and the trajectory prediction of the target being tracked is completed.

[0060] It should be noted that after obtaining the tracking target in each image, the presence of the tracking target in multiple consecutive images can be analyzed to predict the target's position in the future. Therefore, we combine the tracking target identified in multiple images with the motion state of the seeker to analyze the motion state of the tracking target in the current image. Based on this, and combined with existing dynamic models, we complete the prediction of the target's position in the future. Therefore, when predicting images, a three-dimensional coordinate system is first constructed based on the position of the seeker. Then, the position of the tracking target's location in multiple images is analyzed to predict the target's trajectory.

[0061] It should be further explained that when predicting the trajectory of the tracked target, the trajectory can be predicted by the target's current position and real-time velocity. Therefore, when predicting the trajectory of the tracked target, this invention first analyzes the positional changes of the tracked target and the seeker in multiple images to obtain the target's real-time velocity at the current moment, and then predicts the target's trajectory.

[0062] It should be further explained that when predicting the trajectory of the tracked target, the real-time speed of the seeker can be known based on the sensors inside the seeker. Furthermore, the distance between the seeker and the tracked target is inversely proportional to the size of the tracked target in the image captured by the seeker. Therefore, based on the distance the seeker moves between two time points and the size of the tracked target in the image, the distance change of the tracked target between the two time points can be obtained. Since the real-time speed of the tracked target when the first frame of the image is acquired can be obtained using relatively complex existing techniques, the real-time speed of the tracked target at each time point can be obtained based on the distance change of the tracked target between the two time points, and then the trajectory of the tracked target can be predicted.

[0063] Specifically, acquiring the first image captured by the seeker head. The specific formula for calculating the real-time velocity of the target when tracking an image is as follows:

[0064] ;

[0065] In the formula, Indicates the seeker head taking the first shot The real-time speed at which the target is tracked when tracking the target image. Indicates the seeker head taking the first shot The real-time speed at which the target is tracked when tracking the target image. Indicates the seeker head taking the first shot The time it took to capture the image of the target being tracked. Indicates the seeker head taking the first shot The real-time speed of the seeker when tracking a target image. Indicates the seeker head taking the first shot The real-time speed of the seeker when tracking a target image. Indicates the seeker head taking the first shot The time it took to capture the image of the target being tracked. Indicates the first image captured by the seeker. The sum of the areas of all target regions within the image of the target being tracked. Indicates the first image captured by the seeker. The sum of the areas of all target regions within the image of the target being tracked. This represents the sigmoid function, which is used in this embodiment for normalization. This is an exponential function with the natural constant as its base; this embodiment is used to prevent the denominator from being zero. Furthermore, in obtaining... At that time, everything else in the formula is known.

[0066] Among them, in the calculation of the seeker head to capture the first When determining the real-time velocity of the target when capturing the first image, since the real-time velocity of the target when the seeker captures the first image is known, the real-time velocity of the target when the seeker captures the second image can be obtained using the above formula. Then, based on the method for obtaining the real-time velocity of the target when the seeker captures the second image, the real-time velocity of the target when the seeker captures the third image can be obtained using the above formula. This process is repeated iteratively until the real-time velocity of the target when the seeker captures the third image is obtained. The real-time speed at which the target is tracked when tracking an image of the target.

[0067] It should be noted that, To capture the first image of the seeker head Zhang Dao photographed the first The distance the seeker moves when tracking a target image; A positive value indicates that the seeker head captured the first image. The size of the target being tracked in the image is larger than the size captured by the seeker. The size of the target in the image indicates that the distance between the target and the seeker is increasing, based on the characteristic that targets appear larger when closer and smaller when farther away. A negative value indicates that the distance between the tracking target and the seeker is decreasing; therefore, according to right Make corrections; The larger the value, the better. The smaller the value, the better. The larger the value, the better.

[0068] Furthermore, the seeker head will capture the first... When tracking a target image, the real-time velocity of the target minus the velocity captured by the seeker in the first image is used to track the target. The difference between the real-time velocity of the target and the velocity of the target when the seeker captures the first image is denoted as the value of the difference between the real-time velocity of the target and the velocity of the target when the seeker captures the first image. The acceleration of the target when tracking an image of the target.

[0069] Furthermore, a three-dimensional spatial coordinate system is constructed with the position of the seeker when it captures the first image of the tracking target as the origin, and the position and velocity of the seeker when capturing each image of the tracking target are obtained.

[0070] Furthermore, based on the seeker head capturing the first... The position of the seeker head in the three-dimensional coordinate system when tracking target images, based on the first image captured by the seeker head. Using the lower left corner of the target image as the origin, a two-dimensional coordinate system is constructed, and the image captured by the seeker within this two-dimensional coordinate system is obtained. The coordinates of the centroid of a region formed by all target areas within the target image are used in conjunction with the seeker's image capture. When tracking target images, the acceleration and velocity of the target are monitored. Using existing dynamic models, the acceleration and velocity of the target are predicted for the first image captured by the seeker. Track the target's trajectory when tracking an image. Complete the trajectory prediction for the tracked target.

[0071] This concludes the embodiment.

Claims

1. An image sequence analysis method for predicting the dynamic target trajectory of a seeker, characterized in that, include: The process involves acquiring images of the target at multiple angles when it is not in flight, along with a sequence of images of the target captured by the seeker. The real-time speed and position of the seeker are obtained when each target image is acquired, along with the real-time speed of the target when the first target image is acquired. An edge detection algorithm is used to identify regions within each target image. Based on the matching degree between each region in each target image captured by the seeker and the target at multiple angles when it is not in flight, and the difference in matching degree between different regions within the image and the target at multiple angles when it is not in flight, the credibility of each region in each target image captured by the seeker as a target region is determined, thus identifying potential target regions within each target image. Based on the distribution of potential target regions surrounding each potential target region in each target image captured by the seeker, and the difference in credibility between each potential target region and its surrounding potential target regions as target regions, all target regions within each target image are identified. Finally, based on the comparison between each target image captured by the seeker and the previous target image... The real-time speed of the seeker when capturing each tracking target image is determined by the area difference of all target regions within the target image, the real-time speed of the seeker when capturing each tracking target image and the previous tracking target image, and the real-time speed of the tracking target when acquiring the first tracking target image. This allows for trajectory prediction of the tracking target. The specific steps for obtaining the real-time speed of the tracking target when capturing each tracking target image are as follows: Based on the real-time speed of the tracking target when capturing the first tracking target image, the capture time of the first and second tracking target images, the real-time speed of the seeker when capturing the first and second tracking target images, and the sum of the areas of all target regions within the first and second tracking target images, the real-time speed of the tracking target when capturing the second tracking target image is obtained. Following the method for obtaining the real-time speed of the tracking target when capturing the second tracking target image, the real-time speed of the tracking target when capturing the third tracking target image is obtained. This process is repeated continuously to obtain the real-time speed of the tracking target when capturing the third tracking target image. The real-time velocity of the target being tracked when the seeker captures the second image of the target is calculated using the following formula: ； In the formula, Indicates the seeker head taking the first shot The real-time speed at which the target is tracked when tracking the target image. Indicates the seeker head taking the first shot The real-time speed at which the target is tracked when tracking the target image. Indicates the seeker head taking the first shot The time it took to capture the image of the target being tracked. Indicates the seeker head taking the first shot The real-time speed of the seeker when tracking a target image. Indicates the seeker head taking the first shot The real-time speed of the seeker when tracking a target image. Indicates the seeker head taking the first shot The time it took to capture the image of the target being tracked. Indicates the first image captured by the seeker. The sum of the areas of all target regions within the image of the target being tracked. Indicates the first image captured by the seeker. The sum of the areas of all target regions within the image of the target being tracked. This represents the sigmoid function. It is an exponential function with the natural constant as its base.

2. The image sequence analysis method for dynamic target trajectory prediction of a seeker according to claim 1, characterized in that, The specific steps for obtaining the reliability of each region within each tracking target image captured by the seeker as the tracking target region are as follows: Obtain the first image captured by the seeker... The first image within the tracking target The area and the tracked target were in the non-flying state. The maximum matching degree among all images at the nth angle is denoted as the nth image captured by the seeker. The first image within the tracking target The degree of matching between each region and the tracking target; among which, The preset number of angles; the number of images captured by the seeker head. The first image within the tracking target The area and the tracked target were in the non-flying state. The image of the target when it is not in flight, corresponding to the maximum matching degree among all images at all angles, is denoted as the image of the target captured by the seeker. The first image within the tracking target The image corresponding to each region; the first image captured by the seeker head. The first image within the tracking target The area and the first image captured by the guide head The first image within the tracking target The matching degree of the image corresponding to the region is denoted as the first image captured by the seeker. The first image within the tracking target The region is in the first Matching degree based on a region; if the seeker captures the first... The first image within the tracking target The degree of matching between each area and the tracked target, and the first image captured by the seeker. The first image within the tracking target The region is in the first If the absolute value of the difference in matching degree based on each region is less than or equal to 0.1, then the first image captured by the seeker will be... The first image within the tracking target The area is designated as the first area captured by the guidance head. The first image within the tracking target Similar areas in each region; based on the images captured by the seeker head... The first image within the tracking target Similar areas in each region, and the first image captured by the seeker. The first image within the tracking target The degree of matching between each region and the tracked target is used to obtain the first image captured by the seeker. The first image within the tracking target Each region is used to track the credibility of the target region.

3. The image sequence analysis method for dynamic target trajectory prediction of a seeker according to claim 2, characterized in that, The first image captured by the guide head The first image within the tracking target The specific formula for calculating the credibility of the target area for each region is as follows: ； In the formula, Indicates the first image captured by the seeker. The first image within the tracking target Each region is used to track the credibility of the target region. Indicates the first image captured by the seeker. The first image within the tracking target The degree of matching between each region and the tracking target. Indicates the first image captured by the seeker. The number of regions within the target image being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The number of similar regions in each region To prevent hyperparameters with a denominator of 0, This represents the sigmoid function.

4. The image sequence analysis method for dynamic target trajectory prediction of a seeker according to claim 1, characterized in that, The specific steps for obtaining the suspected target region within each tracking target image are as follows: Preset a confidence threshold. If the seeker head captures the first... The first image within the tracking target The credibility of each region as the target region is greater than [a certain value]. Then the first image captured by the seeker head will be... The first image within the tracking target Each area is designated as a suspected target area.

5. The image sequence analysis method for dynamic target trajectory prediction of a seeker according to claim 1, characterized in that, The specific steps for obtaining all target regions within each tracking target image are as follows: Acquire the first image captured by the seeker head... The first image within the tracking target The suspected target area and the first The centroid of a suspected target area; Using the convex hull method, in the first image captured by the seeker... Within the target tracking image, generate a second image that simultaneously contains images captured by the seeker head. The first image within the tracking target The suspected target area and the first The area suspected to be the target area was marked as the first area captured by the seeker head. The first image within the tracking target The suspected target area and the first The connected area of ​​a suspected target region; based on the image captured by the seeker head... The first image within the tracking target The suspected target area and the first The connected area of ​​the suspected target area does not belong to the first... The suspected target area and the first The area of ​​other areas in the suspected target area, the first image captured by the seeker head The first image within the tracking target The suspected target area and the first The area of ​​the connected region of the suspected target area, the first image captured by the seeker. The first image within the tracking target The suspected target area and the first The Euclidean distance of the centroid of the suspected target area, and the first image captured by the seeker. The first image within the tracking target The suspected target area and the first The difference in the credibility of the suspected target area is due to the discrepancy in the reliability of the target area, and the result is the first image captured by the seeker. The first image within the tracking target The suspected target area and the first Each suspected target area contributes to the credibility of the tracked target; based on the images captured by the seeker... The first image within the tracking target The suspected target area and the first Each suspected target area contributes to the credibility of the tracked target, determining the reliability of the image captured by the seeker. The first image within the tracking target The suspected target area and the first Is the suspected target area the target area? 6. The image sequence analysis method for dynamic target trajectory prediction of a seeker according to claim 5, characterized in that, The first image captured by the guide head The first image within the tracking target The suspected target area and the first The specific formula for calculating the confidence level that all suspected target areas are tracking targets is as follows: ; In the formula, Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first Each suspected target area contributes to the credibility of the tracked target. Indicates the first image captured by the seeker. The first image within the tracking target The centroid of the suspected target area and the first The Euclidean distance of the centroid of a suspected target region. Indicates the first image captured by the seeker. The first image within the tracking target The number of suspected target areas is used to assess the credibility of the target area being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The number of suspected target areas is used to assess the credibility of the target area being tracked. Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first The area of ​​the connected regions of a suspected target area. Indicates the first image captured by the seeker. The first image within the tracking target The suspected target area and the first In the connected area of ​​the suspected target area, except for the first The suspected target area and the first The area of ​​other areas outside the suspected target area, To prevent hyperparameters with a denominator of 0, It is an exponential function with the natural constant as its base. This represents the sigmoid function. This represents the absolute value function.

7. The image sequence analysis method for dynamic target trajectory prediction of a seeker according to claim 5, characterized in that, The judgment of the first image captured by the guide head The first image within the tracking target The suspected target area and the first The specific steps for determining whether a suspected target area is indeed a target area are as follows: Preset a confidence threshold. If the seeker head captures the first... The first image within the tracking target The suspected target area and the first All suspected target areas have a confidence level greater than the confidence threshold for tracking targets. Then the first image captured by the seeker head will be... The first image within the tracking target The suspected target area and the first All suspected target areas are recorded as target areas.

8. The image sequence analysis method for dynamic target trajectory prediction of a seeker according to claim 1, characterized in that, The specific steps for completing the trajectory prediction of the tracked target are as follows: The seeker head takes a picture of the... When tracking a target image, the real-time velocity of the target minus the velocity captured by the seeker in the first image is used to track the target. The difference between the real-time velocity of the target and the velocity of the target when the seeker captures the first image is denoted as the value of the difference between the real-time velocity of the target and the velocity of the target when the seeker captures the first image. The acceleration of the target is measured when the first image of the target is captured; a three-dimensional coordinate system is constructed with the position of the seeker when the first image of the target is captured as the origin, and the position and velocity of the seeker are obtained when each image of the target is captured; based on the image captured by the seeker... The position of the seeker head in the three-dimensional coordinate system when tracking target images, based on the first image captured by the seeker head. Using the lower left corner of the target image as the origin, a two-dimensional coordinate system is constructed, and the image captured by the seeker within this two-dimensional coordinate system is obtained. The coordinates of the centroid of a region formed by all target areas within the target image are used in conjunction with the seeker's image capture. When tracking a target image, the acceleration and velocity of the target are monitored. Using existing dynamic models, the acceleration and velocity of the target are predicted for the first image captured by the seeker. Tracking the target's trajectory when tracking the target image.

Images