Image recognition-based target point observation method, system, and storage medium
By using an image recognition-based method to generate star trails through short-exposure continuous shooting, and calculating the trajectory lines and velocities, the motion recognition problem in multi-target situations is solved, and high-precision spatial target prediction is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING KAIYUN PARALLEL SPACE TECH CO LTD
- Filing Date
- 2023-01-10
- Publication Date
- 2026-05-22
Smart Images
Figure CN116228813B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of optical image recognition technology for space objects, and specifically to a target point observation method, system, and storage medium based on image recognition. Background Technology
[0002] For operations requiring orbital parameters of space targets, such as rocket launches, collision warnings, and transit forecasts, having an independent, controllable, and reliable data source is essential. Currently, monitoring methods for space targets (satellites, rocket bodies, debris, etc.) mainly include radar monitoring and visible light observation. Radar is complex to deploy and costly, but it has high sensitivity to low-Earth orbit targets and can observe targets smaller than 10cm; it is primarily used by government agencies. Optical observation mainly uses astronomical telescopes; optical telescopes are flexible to deploy and low-cost, making them the most commonly used method by businesses and individuals.
[0003] The monitoring of space targets by optical telescopes is mainly divided into tracking and surveying. Tracking involves scheduling cameras to take pictures according to their positions based on provided TLE parameters. Surveying involves pointing the telescope in a specified direction and recording satellites passing within the field of view for orbit matching and updates. In survey mode, the main workflow is to remove background satellites by overlaying the captured images and setting thresholds. The remaining satellite points constitute the trajectory point set of the space target. By connecting the trajectories, the satellite's direction and velocity can be obtained, allowing prediction of the target's pixel position in the next image or its position within a certain future timeframe. Due to the short observation time, the accuracy is relatively low, and the error rate increases for predictions made over longer periods.
[0004] Currently used methods primarily address the scenario where there is only one spatial target in the field of view. These methods mainly involve collecting the coordinates of the spatial target from multiple processed images and fitting this point set to obtain the target's motion direction and approximate velocity. However, when multiple targets are moving within the field of view, the resolved coordinate set becomes irregular, leading to the inability to determine the target's trajectory or to obtain incorrect direction or velocity. Consequently, the results become unreliable, impacting the prediction of the spatial target. Summary of the Invention
[0005] Technical Objective: To address the aforementioned technical problems, this invention proposes a target point observation method, system, and readable storage medium based on image recognition. Based on the image recognition processing method and the continuity of star trails, it can correctly identify the motion direction and velocity of each space target when multiple space targets exist in the field of view.
[0006] Technical solution: To achieve the above technical objectives, the present invention adopts the following technical solution:
[0007] A target point observation method based on image recognition, characterized by comprising the following steps:
[0008] Using an optical telescope with a camera function, the space target is photographed continuously with short exposures to obtain multiple raw images, in which the space target forms star trails in the raw images.
[0009] The acquired original images are processed separately to obtain the corresponding star trail images;
[0010] In the same star trail image, the two endpoints of the trajectory lines of each star trail are taken, and a straight line is determined through the two endpoints. The velocity is calculated based on the exposure time of the original image and the distance between the two endpoints. The straight line and the velocity are respectively identified as the motion trajectory and motion velocity of the space target corresponding to the star trail.
[0011] Compare and match the motion trajectory lines and motion speeds of star trails in each star trail image. If the trajectory lines of two or more star trails are on the same straight line and their motion speeds are equal or the difference in motion speeds is within a preset range, the corresponding star trails are identified as the same space target.
[0012] For star trail images identified as belonging to the same space target, rays are drawn from the endpoints of the earlier star trails to the endpoints of the later star trails, according to the chronological order in which the original images were captured. The direction of these rays is determined as the direction of motion of the space target.
[0013] Preferably, after determining the direction of motion of the space targets, predicting the location where each space target will appear next includes the following steps:
[0014] Based on the changes in time, determine the starting point of the motion direction of any spatial target, and combine the corresponding spatial target's motion speed and direction to predict the location where the corresponding spatial target will appear next.
[0015] Preferably, the optical telescope has a camera resolution of m*n, the spatial target moves along a direction parallel to the m-resolution side, the angle at which the spatial target is identified is β, and the exposure time satisfies the requirement of equation (1):
[0016]
[0017] Where t represents the exposure time, α represents the field of view of the camera, d is the diameter of the star point representing the space target in the obtained original image, 2d represents the identifiable length of the star trail formed after the space target moves, and ω represents the angular velocity of the space target.
[0018] Preferably, the acquired multiple original images are processed separately to obtain corresponding star trail images, including the following steps:
[0019] Denoise reduction processing is performed on the original image;
[0020] The original image after noise reduction is overlaid and thresholded to identify background stars in the image and remove them from the image.
[0021] The image with stars removed is filtered according to a brightness threshold to obtain a star trail image.
[0022] A target point observation system based on image recognition, characterized in that it includes:
[0023] The raw image acquisition device is used to continuously capture images of space targets with short exposures to obtain multiple raw images, in which the space targets form star trails in the raw images.
[0024] The image processing module is used to process multiple acquired raw images separately to obtain corresponding star trail images;
[0025] The calculation module takes the two endpoints of the trajectory lines of each star trail in the same star trail image, determines a straight line through the two endpoints, and calculates the velocity based on the exposure time of the original image and the distance between the two endpoints. The straight line and the velocity are respectively identified as the motion trajectory and motion velocity of the space target corresponding to the star trail.
[0026] The identification module compares and matches the motion trajectory lines and motion speeds of star trails in each star trail image. If the trajectory lines of two or more star trails are on the same straight line and their motion speeds are equal or the difference in motion speeds is within a preset range, the corresponding star trails are identified as the same space target.
[0027] The direction confirmation module is used to draw rays from the endpoints of the star trails that are identified as the same space target to the endpoints of the star trails that were captured earlier, according to the chronological order of the original images. The direction of the rays is determined as the direction of motion of the space target.
[0028] Preferably, it further includes: a prediction module, used to determine the starting point of the motion direction of any space target based on the change in time, and predict the position of the next appearance of the corresponding space target by combining the motion speed and motion direction of the corresponding space target.
[0029] A storage medium, characterized in that a computer program is stored on the storage medium, and the computer program implements the method when executed by a processor.
[0030] Beneficial effects: Due to the adoption of the above technical solution, the present invention has the following beneficial effects:
[0031] This invention proposes a target point observation method based on image recognition. By using short exposure, short star trails are formed in multiple original images. Based on the continuity of the star trails, the spatial target and its direction and speed of motion are determined. Moreover, the starting point of the spatial target's direction of motion is determined according to the change of time points, which can predict the approximate location of the next appearance of the spatial target. Attached Figure Description
[0032] Figure 1 This is a flowchart of the target point observation method based on image recognition proposed in this invention;
[0033] Figure 2 A schematic diagram of star points and star trails;
[0034] Figure 3 This is a schematic diagram showing the camera's resolution and the direction of star trail movement.
[0035] Figure 4 This is a schematic diagram showing the relationship between the field of view angle and the star trail angle.
[0036] Figure 5 A schematic diagram of star points exposed in a short time;
[0037] Figure 6 A schematic diagram of star trails captured in a short exposure;
[0038] Figure 7 The image before brightness threshold filtering;
[0039] Figure 8 The image after brightness threshold filtering;
[0040] Figure 9 A schematic diagram of the predicted trajectory for a space target;
[0041] Figure 10 This is the first image of star trails;
[0042] Figure 11 This is the second image of star trails;
[0043] Figure 12 This is a schematic diagram for determining the direction of movement of a space target based on star trail images. Detailed Implementation
[0044] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
[0045] like Figure 1 As shown, this invention proposes a target point observation method based on image recognition, including the following steps:
[0046] Using an optical telescope with a camera function, the space target is photographed continuously with short exposures to obtain multiple raw images, in which the space target forms star trails in the raw images.
[0047] The acquired original images are processed separately to obtain the corresponding star trail images;
[0048] In the same star trail image, the two endpoints of the trajectory lines of each star trail are taken, and a straight line is determined through the two endpoints. The velocity is calculated based on the exposure time of the original image and the distance between the two endpoints. The straight line and the velocity are respectively identified as the motion trajectory and motion velocity of the space target corresponding to the star trail.
[0049] Compare and match the motion trajectory lines and motion speeds of star trails in each star trail image. If the trajectory lines of two or more star trails are on the same straight line and their motion speeds are equal or the difference in motion speeds is within a preset range, the corresponding star trails are identified as the same space target.
[0050] For star trail images identified as belonging to the same space target, rays are drawn from the endpoints of the earlier star trails to the endpoints of the later star trails, according to the chronological order in which the original images were captured. The direction of these rays is determined as the direction of motion of the space target.
[0051] Preferably, after determining the direction of motion of the space targets, predicting the location where each space target will appear next includes the following steps:
[0052] Based on the changes in time, determine the starting point of the motion direction of any spatial target, and combine the corresponding spatial target's motion speed and direction to predict the location where the corresponding spatial target will appear next.
[0053] The target observation method based on image recognition proposed in this invention first uses an optical telescope with camera functionality to take short exposures of the space target, forming the image of star trails from the target's stars. Then, by recognizing the motion of the star trails, the length and direction of the trails are determined. This same process is then performed on multiple images. Finally, the obtained star trail velocities and directions are matched to obtain the approximate direction of motion of the space target, predicting the location where the target point will appear next.
[0054] Combination Figures 1 to 12 The detailed steps of this invention are as follows:
[0055] 1. Configure the camera exposure time parameters.
[0056] Because targets at different altitudes move at different speeds—lower orbit targets have faster angular velocities, while higher orbit targets have slower angular velocities—the minimum exposure time is set to be sufficient to identify the target's trajectory. The pixel size corresponding to a star point on the target is roughly calculated based on the camera resolution, and then the diameter of the star point is obtained.
[0057] Assumption:
[0058] The diameter of the target star in space is d, and the recognizable length of the star trail formed after the target moves is 2d. Figure 1 As shown.
[0059] The camera's resolution is m*n, such as Figure 2 As shown. The camera's field of view is α, as... Figure 3 As shown.
[0060] If the space target moves along a direction parallel to the m-resolution edge, then the angle β that the space target needs to be identified is:
[0061]
[0062] The angular velocity of the space target is ω;
[0063] The exposure time is:
[0064]
[0065] A stationary star in space, such as Figure 4 As shown, star trails formed after a short exposure, such as Figure 5 As shown.
[0066] 2. Imaging. When photographing satellites, the camera takes continuous shots, capturing only one image at a time. The shooting interval is very short, within a few seconds. The movement of space targets within this short timeframe can be considered essentially along a straight line.
[0067] 3. Star trail image recognition
[0068] Image processing techniques such as median filtering are used to reduce electronic noise in the image. Background stars are identified through image overlay, thresholding, or other methods. Overlaying the background stars onto the original image eliminates the background, yielding the image of the space target. In other words, after identifying the coordinates of the stars, subtracting them from the original image leaves the star trails. Filtering by the brightness threshold of nearby points yields a complete star trail image, such as... Figure 6 and Figure 7 The images shown are star trail images before and after brightness threshold filtering.
[0069] 4. Calculation of the direction and velocity of star trails
[0070] By connecting the pixel positions at both ends of the star track, a short-term trajectory line of the satellite can be obtained.
[0071]
[0072] Where (x1, y1) and (x2, y2) are... Figure 7 The two endpoints (start point and end point) of the processed star trail.
[0073] Based on the length of the star trails and the exposure time of the shot, the speed of the moving target in space can be calculated, and the approximate coordinates of the star can be obtained based on the time of the next shot.
[0074] With only one image, we can obtain the trajectory of the moving object in space and its next possible location (one of the two), such as... Figure 8 The lines shown represent the trajectory of a spatial target. Points A and B on the lines represent possible locations for the next shot, but their directions cannot be determined.
[0075] 5. Analyze multiple images, merge the results, and obtain the orientation.
[0076] Figure 9 This is the first image of star trails. Figure 10 This is the second star trail image. By processing multiple star trail images, multiple coordinate points of star trails can be obtained. A ray is drawn from the coordinate points of the first star trail to the coordinate points of the second star trail; the direction of this ray represents the direction of movement of the space target. For example... Figure 11 As shown by the middle arrow, this ray is formed by connecting the points where the images were taken (first image to second image) to obtain the ray.
[0077] 6. Elimination of interference from multiple satellite orbits
[0078] During the shooting process, we encountered a situation with multiple star trails. Since we knew the direction and speed of the required star trails, and the direction, size or position of the star trails was different from the current star trails, we were able to distinguish each star trail.
[0079] This invention proposes a target point observation method based on image recognition. Through short exposures, short star trails are formed in the image. The continuity of the star trails effectively avoids interference between target point sets. During processing, the start and end points of the star trails are identified, and a straight line is determined between these two points. This straight line represents the trajectory of the space target. The velocity of the space target can be calculated by determining the camera's exposure time and the distance between the start and end points. With only one image, it's impossible to determine which of the two endpoints of the star trail is the start and end point. Multiple images are required. By comparing and matching the trajectory lines and velocities of the space target across multiple images, when the trajectories lie on the same straight line and the velocities are approximately equal, they can be considered the same space target. Furthermore, by determining the starting point of the space target's motion direction based on changes in time, the approximate location of the next appearance of the space target can be predicted, achieving both observation and prediction of space targets.
[0080] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any way, and all technical solutions obtained by equivalent substitution or equivalent transformation fall within the protection scope of the present invention.
Claims
1. A target point observation method based on image recognition, characterized in that, Including the following steps: Using an optical telescope with a camera function, the space target is photographed continuously with short exposures to obtain multiple raw images, in which the space target forms star trails in the raw images. The acquired original images are processed separately to obtain the corresponding star trail images; In the same star trail image, the two endpoints of the trajectory lines of each star trail are taken, and a straight line is determined through the two endpoints. The velocity is calculated based on the exposure time of the original image and the distance between the two endpoints. The straight line and the velocity are respectively identified as the motion trajectory and motion velocity of the space target corresponding to the star trail. Compare and match the motion trajectory lines and motion speeds of star trails in each star trail image. If the trajectory lines of two or more star trails are on the same straight line and their motion speeds are equal or the difference in motion speeds is within a preset range, the corresponding star trails are identified as the same space target. For star trail images that are identified as the same space target, rays are drawn from the endpoint of the earlier star trail to the endpoint of the later star trail, according to the chronological order of the original image capture time. The direction of the ray is determined as the direction of motion of the space target. The process involves processing multiple original images to obtain corresponding star trail images, including the following steps: Denoising the original image; The original image after noise reduction is overlaid and thresholded to identify background stars in the image and remove them from the image. The image with stars removed is filtered according to a brightness threshold to obtain a star trail image.
2. The target point observation method based on image recognition according to claim 1, characterized in that, After determining the direction of motion of space targets, the next location of each space target is predicted, including the following steps: Based on the changes in time, determine the starting point of the motion direction of any spatial target, and combine the corresponding spatial target's motion speed and direction to predict the location where the corresponding spatial target will appear next.
3. The target point observation method based on image recognition according to claim 1, characterized in that: The camera resolution of the optical telescope is Space target along parallel The angle at which spatial targets are identified is determined by the motion along the resolution side. The exposure time satisfies the requirement of equation (1): in, Indicates the exposure time. Indicates the camera's field of view. The diameter of the star representing the space target in the obtained original image is , This indicates the identifiable length of the star trail formed after a space target moves. This represents the angular velocity of a space target.
4. A target point observation system based on image recognition, characterized in that, include: The raw image acquisition device is used to continuously capture images of space targets using short exposures to obtain multiple raw images, in which the space targets form star trails in the raw images. The image processing module is used to process multiple acquired raw images separately to obtain corresponding star trail images; The calculation module takes the two endpoints of the trajectory lines of each star trail in the same star trail image, determines a straight line through the two endpoints, and calculates the velocity based on the exposure time of the original image and the distance between the two endpoints. The straight line and the velocity are respectively identified as the motion trajectory and motion velocity of the space target corresponding to the star trail. The identification module compares and matches the motion trajectory lines and motion speeds of star trails in each star trail image. If the trajectory lines of two or more star trails are on the same straight line and their motion speeds are equal or the difference in motion speeds is within a preset range, the corresponding star trails are identified as the same space target. The direction confirmation module is used to draw rays from the endpoint of the star track that is identified as the same space target to the endpoint of the star track that was captured earlier, according to the chronological order of the original images. The direction of the rays is determined as the direction of motion of the space target. The image processing module processes multiple acquired original images to obtain corresponding star trail images, including the following steps: Denoising the original image; The original image after noise reduction is overlaid and thresholded to identify background stars in the image and remove them from the image. The image with stars removed is filtered according to a brightness threshold to obtain a star trail image.
5. A target point observation system based on image recognition according to claim 4, characterized in that, Also includes: The prediction module is used to determine the starting point of the motion direction of any space target based on the changes in time, and predict the position of the next appearance of the corresponding space target by combining the motion speed and motion direction of the corresponding space target.
6. A storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 3.