Spacecraft Parachute Intelligent Tracking System and Method Based on Vehicle Theodolite

Abstract

The present invention relates to the technical field of target tracking, and particularly to an intelligent tracking system and method for a spacecraft parachute based on a vehicle-mounted theodolite, including a vehicle-mounted theodolite, an external transmission device, an industrial control computer, a video capture card, an image processing module, and an image bit conversion processing algorithm module for converting the image bits. The vehicle-mounted theodolite is connected to the industrial control computer, and the video capture card is connected to the industrial control computer, the vehicle-mounted theodolite, and the external transmission device. The image processing module and the image bit conversion processing algorithm module are configured on the industrial control computer, and the image processing module and the image bit conversion processing algorithm module can be run through the industrial control computer. The present invention solves the problems of deficiencies in light dependence, target misidentification, environmental adaptability, real-time processing ability, and feature learning of the existing parachute tracking system. The system has high tracking accuracy and strong robustness, and significantly improves the safety and reliability during the spacecraft landing process.

Description

Technical Field

[0001] The present invention belongs to the technical field of target tracking, and particularly relates to an intelligent tracking system and method for a spacecraft parachute based on a vehicle-mounted theodolite. Background Art

[0002] During the landing process of a spacecraft, the deployment and tracking of the parachute are important links to ensure the safe return of the spacecraft. However, existing parachute tracking algorithms and systems are prone to problems such as target loss or false tracking under complex environmental conditions, such as strong winds and light changes. Especially when the parachute is similar in shape to the return capsule, it leads to a decrease in tracking accuracy and affects the safe landing of the spacecraft.

[0003] Existing parachute tracking systems mainly rely on visible light cameras or infrared cameras for image acquisition, and use existing target tracking algorithms such as Kalman filtering, Meanshift, Camshift, etc. to achieve the tracking of the parachute. However, these target tracking algorithms have many deficiencies in practical applications. First of all, strong light dependence is a significant problem. The performance of visible light cameras drops significantly under low light or backlight conditions, resulting in unstable tracking effects and difficulty in maintaining high-precision tracking in various lighting environments. Secondly, the problem of target misidentification is particularly prominent. Since the parachute and the return capsule are similar in shape at certain angles, existing target tracking algorithms are prone to misidentifying the return capsule as the parachute, leading to false tracking and thus affecting the overall reliability and safety of the tracking system. In addition, existing target tracking algorithms have poor robustness to occlusion and partial target loss. When the parachute is partially occluded by other objects during movement, existing target tracking algorithms are difficult to correctly handle the re-detection and recovery after target loss, resulting in frequent tracking interruptions or false tracking. Environmental adaptability is also a key issue. Under complex environmental conditions such as strong winds and heavy fog, the accuracy of existing target tracking algorithms is significantly reduced, making it difficult to meet the requirements for high reliability in practical applications. More seriously, existing target tracking algorithms lack the ability of deep feature learning. They usually rely on manually designed features such as color histograms, HOG (Histogram of Oriented Gradient), etc., without using deep learning methods to automatically learn more advanced features. This makes it difficult for existing target tracking algorithms to achieve sufficient discrimination and robustness in complex scenarios, especially in the case of the complex appearance and dynamic behavior of the parachute. At the same time, existing target tracking systems and algorithms have insufficient real-time processing capabilities when dealing with high-frame-rate image data, making it difficult to meet the efficient tracking requirements of more than 30fps, resulting in frame loss and unstable tracking problems.

[0004] There are also some existing target tracking systems that use dual cameras (visible light and infrared) and combine image fusion technology to improve tracking performance under different lighting conditions. At the same time, deep learning algorithms (such as YOLOv5, YOLOv8) are used for target detection to enhance the robustness and accuracy of the system. Although these systems solve the problems of light dependence and mis-identification to a certain extent, they still have the following disadvantages: (1) High system complexity: The data fusion of dual cameras increases the system complexity and cost. (2) Low processing efficiency: Even with the use of deep learning algorithms, the real-time processing speed is still difficult to meet the high frame rate requirements. (3) Insufficient adaptability to specific environments: In extreme environments, such as under the influence of strong winds, the accuracy of target recognition and tracking is still limited. (4) Insufficient mechanism for correcting mis-tracking: There is a lack of effective strategies to correct mis-tracking, especially in the case of complex morphological changes. Summary of the Invention

[0005] In view of this, the present invention aims to provide an intelligent tracking system and method for spacecraft parachutes based on a vehicle-mounted theodolite to improve the accuracy, real-time performance, and robustness of tracking.

[0006] To achieve the above object, the technical solution of the present invention is realized as follows:

[0007] On the one hand, the present invention provides an intelligent tracking system for spacecraft parachutes based on a vehicle-mounted theodolite, including a vehicle-mounted theodolite, an external transmission device, an industrial control computer, a video capture card, an image processing module, and an image bit conversion processing algorithm module for converting the image bit number. The vehicle-mounted theodolite is connected to the industrial control computer, and the video capture card is connected to the industrial control computer, the vehicle-mounted theodolite, and the external transmission device. The image processing module and the image bit conversion processing algorithm module are configured on the industrial control computer, and the image processing module and the image bit conversion processing algorithm module can be run through the industrial control computer. Among them, the vehicle-mounted theodolite includes a theodolite body and a multi-spectral imaging device, and the multi-spectral imaging device is integrally installed on the theodolite body to collect image data and position data of the parachute in real time. The industrial control computer includes a processor, a memory, a graphics processor, a display device, and a storage module to perform real-time processing on the image data, display the tracking result in real time, and store the data.

[0008] Furthermore, the image processing module includes an OSTrack tracking algorithm module and a YOLOv8 target detection algorithm module to perform real-time tracking and auxiliary detection on the parachute. When the OSTrack tracking algorithm module loses the target, the YOLOv8 target detection algorithm module serves as an auxiliary decision-making tool, responsible for re-detecting the parachute and transmitting the detection result to the OSTrack tracking algorithm module to re-initialize the OSTrack tracking algorithm module.

[0009] Furthermore, the OSTrack tracking algorithm module uses the ViT-based single-stream object tracking algorithm OSTrack as the tracking algorithm, and the following optimizations and extensions are made:

[0010] Joint feature learning and relationship modeling. By splicing the template image and the search region image and inputting them into the ViT encoder, the self-attention mechanism is used to distinguish the target from the background during the encoding stage, effectively generating discriminative target features;

[0011] Early candidate elimination module. An early candidate elimination module is inserted into the multi-layer encoder of ViT. Using the similarity score between the template image and the search region image, background region candidates are dynamically eliminated;

[0012] False tracking correction. Combining the inverted triangle shape features of the parachute and the spacecraft and the morphological changes under the influence of wind, through shape matching and feature verification strategies, the parachute is correctly tracked and the spacecraft is prevented from being falsely tracked.

[0013] Furthermore, the YOLOv8 object detection algorithm module is trained using its own dataset and the weights are optimized to accurately identify the parachute.

[0014] Furthermore, the multispectral imaging device includes a visible light camera and an infrared camera; the visible light camera transmits image data to the video capture card through the SDI interface method. After the video capture card receives and processes the visible light image data, it is transmitted to the industrial control computer. After being processed by the image processing module configured on the industrial control computer, it is then transmitted to the external device through the video capture card; the infrared camera is connected to the industrial control computer through RTSP. The image data collected by the infrared camera is transmitted to the industrial control computer through RTSP. After being processed by the image bit conversion processing algorithm module and the image processing module configured on the industrial control computer, it is then transmitted to the external device through the video capture card.

[0015] On the other hand, the present invention provides an intelligent tracking method for a spacecraft parachute based on a vehicle-mounted theodolite. Based on the above-mentioned intelligent tracking system for a spacecraft parachute based on a vehicle-mounted theodolite, it includes:

[0016] S1: Image acquisition: The vehicle-mounted theodolite acquires the image data of the parachute through the multispectral imaging device. The visible light image data collected by the multispectral imaging device is transmitted to the industrial control computer through the video capture card using the SDI interface method, and the infrared image data collected by the multispectral imaging device is transmitted to the industrial control computer through RTSP;

[0017] S2: Image preprocessing: The visible light image data is directly input into the image processing module configured on the industrial control computer through the connection method of transmitting the SDI signal to the video capture card. The visible light image is preliminarily processed and directly used by the OSTrack tracking algorithm module; the 14-bit high-order image data collected by the infrared camera is processed by the image bit conversion processing algorithm module and converted into an 8-bit image format; the frame rate of the image data is identified. When the frame rate of the image data exceeds 50fps, the frame extraction method is adopted.

[0018] S3: Target tracking: Through the OSTrack tracking algorithm module, the parachute is tracked in real time by combining the shape characteristics of the parachute and the spacecraft; the intelligent tracking system continuously monitors the size change of the parachute. If the size of the parachute changes abnormally, that is, the area of the parachute is greater than or equal to 2 times the area before the change, or less than or equal to 0.5 times the area before the change, it is determined that the target is lost and transferred to step S4.

[0019] S4: Auxiliary detection: The intelligent tracking system calls the YOLOv8 target detection algorithm module to perform target detection on the current frame image, identify the position of the parachute, use this position as the new tracking target, re-initialize the OSTrack tracking algorithm module, and achieve continuous tracking; otherwise, it is confirmed as a mis-tracked target, and the intelligent tracking system ignores this target and continues to call the YOLOv8 target detection algorithm module for re-detection until the position of the parachute is accurately identified.

[0020] Further, in step S1, the visible light camera of the multispectral imaging device is responsible for capturing clear images of the parachute during the day and under good lighting conditions, and the infrared camera of the multispectral imaging device provides stable image capture capabilities under low light or night conditions to capture infrared images of the parachute.

[0021] Further, step S3 includes the following steps:

[0022] S31: Initialization: Manually select the initial position of the parachute through the user interface of the industrial control computer display device, or automatically detect the position of the parachute based on the vehicle-mounted theodolite's spacecraft parachute intelligent tracking system, and initialize the OSTrack tracking algorithm module.

[0023] S32: Real-time tracking and target loss detection: The parachute is tracked in real time through the OSTrack tracking algorithm module, and the bounding box coordinates of the parachute are output; the intelligent tracking system continuously monitors the size change of the parachute. If the size of the parachute changes abnormally, that is, the area of the parachute is greater than or equal to 2 times the area before the change, or less than or equal to 0.5 times the area before the change, it is determined that the target is lost and transferred to step S4.

[0024] Furthermore, step S4 includes: when using the YOLOv8 object detection algorithm module to perform object detection on the current frame image to identify the position of the parachute, an incorrect tracking correction strategy is adopted. Combining the inverted triangle shape features of the parachute and the spacecraft and the morphological changes under the influence of wind, through shape matching and feature verification, to ensure correct tracking of the parachute and avoid incorrect tracking of the return capsule. Specifically, it includes the following sub-steps:

[0025] S41: Shape feature extraction: Using image processing techniques, through edge detection and contour analysis, extract the shape features of each object detected by the YOLOv8 object detection algorithm module;

[0026] S42: Shape matching and verification: Match the shape features obtained in step S41 with the inverted triangle shape features formed by the preset combination of the parachute and the spacecraft, and calculate the shape similarity score; Compare the shape similarity score with the preset shape similarity score threshold. If the shape similarity score is higher than the threshold, confirm that the object is a combination of the parachute and the return capsule, use the position of this object as the new tracking target, re-initialize the OSTrack tracking algorithm module, and realize continuous tracking; Otherwise, confirm it as an incorrect tracking target, and the intelligent tracking system ignores this target and continues to call the YOLOv8 object detection algorithm module for re-detection until the position of the parachute is accurately identified; Among them, according to the morphological changes of the parachute, the preset shape similarity score threshold is dynamically adjusted.

[0027] Furthermore, the intelligent tracking system stores the tracking data during the tracking process in the storage module of the industrial control computer; The user can view the real-time position of the parachute through the display device; The user can manually select the tracking target, adjust or reset the tracking system, and view the tracking status and tracking system performance at any time.

[0028] Compared with the prior art, the present invention can achieve the following beneficial effects: The present invention supports the RTSP protocol image pulling function, meets the access requirements of multi-source video streams, and has the function of converting 14-bit infrared images to 8-bit display, improving the image processing and display effects. The present invention uses the OSTrack tracking algorithm as the main tracking algorithm and introduces the YOLOv8 object detection algorithm as an auxiliary decision-making tool. When the tracking target is lost, the YOLOv8 object detection algorithm can be called to re-detect and identify the parachute in a timely manner, ensuring the continuity, accuracy, and real-time performance of tracking. The present invention has high real-time processing capabilities. Utilizing a high-performance industrial computer and GPU acceleration, it achieves a real-time processing speed of over 50fps, meeting the requirements of high-frame-rate tasks, and ensuring the stability and tracking accuracy of the tracking system through the frame extraction method. In addition, the present invention combines an anti-mis-tracking correction strategy. Through shape matching and feature verification, it ensures that the tracking system can accurately distinguish the parachute from the return capsule, avoiding mis-tracking the return capsule under the influence of wind, and further enhancing the robustness and safety of the tracking system in complex environments. The system of the present invention has a high degree of integration and supports two connection methods simultaneously, ensuring the efficient synchronization and transmission of camera image data of different types, achieving the efficient transmission and synchronous processing of visible light image data and infrared image data, and reducing the complexity and maintenance cost of the system. The tracking system of the present invention can achieve fully automatic tracking, and the tracking data obtained after tracking (including the bounding box coordinates of the target, miss distance, etc.) can be effectively used for downstream tasks, further enhancing the application value and engineering feasibility of the tracking system. The present invention balances the real-time performance and accuracy of the tracking system. Through the early candidate elimination module and efficient algorithm design, while ensuring tracking accuracy, it improves the real-time processing capabilities of the tracking system and adapts to various application scenarios. The present invention solves the problems of the existing parachute tracking system in terms of light dependence, target mis-identification, environmental adaptability, real-time processing capabilities, and feature learning, and provides an intelligent tracking system and tracking method with high precision, strong robustness, and high real-time processing capabilities, significantly enhancing the safety and reliability during the spacecraft landing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings that form a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0030] Figure 1 It is a schematic structural diagram of the intelligent tracking system for spacecraft parachutes based on vehicle theodolites according to the embodiments of the present invention;

[0031] Figure 2 It is a schematic flowchart of the intelligent tracking method for spacecraft parachutes based on vehicle theodolites according to the embodiments of the present invention;

[0032] Figure 3 Schematic diagram of the mis - tracking correction strategy process according to the embodiments of the present invention Detailed implementation manners

[0033] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. Similar elements in different embodiments are labeled with related similar element numbers. In the following embodiments, many detailed descriptions are provided to enable a better understanding of the present invention. However, those skilled in the art can easily recognize that some of the features can be omitted in different situations, or can be replaced by other elements, materials, or methods. In some cases, some operations related to the present invention are not shown or described in the specification, in order to avoid the core part of the present invention being overwhelmed by excessive descriptions. For those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations based on the descriptions in the specification and the general technical knowledge in the art.

[0034] It should be noted that, without conflict, the embodiments and features in the embodiments of the present invention can be combined with each other to form various embodiments. At the same time, the steps or actions in the method description can also be reordered or adjusted in an obvious manner by those skilled in the art. Therefore, the various sequences in the specification and drawings are only for clearly describing a certain embodiment, and do not mean that they are the necessary sequences, unless it is stated that a certain sequence must be followed.

[0035] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. These are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of the present invention. In addition, the terms "first", "second", etc. are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined with "first", "second", etc. may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise stated, the meaning of "a plurality" is two or more.

[0036] In the description of the present invention, it should be noted that, unless otherwise clearly specified and defined, the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the communication inside two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0037] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0038] As Figures 1 to 3 shown, on the one hand, the present invention provides a spacecraft parachute intelligent tracking system based on a vehicle-mounted theodolite, including: a vehicle-mounted theodolite, an external delivery device, an industrial control computer, an image processing module, an image bit conversion processing algorithm module, and a DeckLink video capture card. The DeckLink video capture card is plugged into the industrial control computer and is connected to the vehicle-mounted theodolite and the external delivery device. The vehicle-mounted theodolite is connected to the industrial control computer, and the industrial control computer is connected to the external delivery device through the DeckLink video capture card. The image processing module and the image bit conversion processing algorithm module are configured on the industrial control computer, and the image processing module can be run through the industrial control computer, wherein:

[0039] Vehicle-mounted theodolite: It includes a theodolite body and a multi-spectral imaging device. The multi-spectral imaging device is integrally installed on the theodolite body and is used to collect image data and position data of the parachute as a tracking target in real time. In some embodiments, the multi-spectral imaging device includes a visible light camera and an infrared camera.

[0040] Industrial control computer: It includes a processor, memory, graphics processor, display device, and storage module to perform real-time processing of image data, display the tracking results in real time, and store the data for subsequent analysis and backtracking. In some embodiments, the processor is selected as Intel Core i9-14900k, providing powerful multi-core processing capabilities to meet the high-frame-rate real-time processing requirements; the memory is selected as 128GB DDR4 memory to ensure the smooth operation of multitasking and large data streams; the graphics processor is selected as NVIDIA GeForce RTX 3080Ti GPU, supporting CUDA acceleration, which can improve the inference speed of deep learning models; a 4TB NVMe solid-state drive is selected to provide fast data read and write capabilities to meet the storage requirements of large-capacity video image data; it is equipped with more than 3 PCIe slots and 1 PCIe X8 slot to support multi-graphics card and high-speed data transfer requirements; a dual-network card design supports high-speed data transfer and network redundancy. With a high-performance industrial control computer and GPU acceleration, a real-time processing speed of more than 50fps can be achieved. When the present invention exceeds 50fps, a frame extraction method is adopted, that is, tracking processing is performed every certain number of frames to balance the system load and tracking accuracy, ensure system stability and tracking accuracy, and maintain real-time performance at the same time.

[0041] Image processing module: It includes an OSTrack tracking algorithm module and a YOLOv8 object detection algorithm module to perform real-time tracking and auxiliary detection of the parachute.

[0042] OSTrack is a single-stream object tracking algorithm based on Vision Transformer (ViT, a neural network architecture based on self-attention mechanism), featuring high precision and high efficiency. The OSTrack tracking algorithm module of the present invention adopts OSTrack as the tracking algorithm and has been optimized and extended in the following aspects: Joint feature learning and relation modeling, by splicing the template image and the search area image and inputting them into the ViT encoder, using the self-attention mechanism to distinguish the target from the background during the encoding stage, and effectively generating discriminative target features; Early candidate elimination module, inserting an early candidate elimination module in the multi-layer encoder of ViT, using the similarity score between the template image and the search area image to dynamically eliminate background area candidates, reducing the computational burden and improving the inference speed; False tracking correction, combining the shape features of the parachute and the return capsule (inverted triangle shape) and the morphological changes under the influence of wind, and ensuring the correct tracking of the parachute and avoiding false tracking of the return capsule through shape matching and feature verification strategies. Among them, the acquisition of the similarity score belongs to the prior art, and reference can be made to "Joint Feature Learning and Relation Modeling for Tracking: A One-Stream Framework, ECCV, 2022".

[0043] The YOLOv8 object detection algorithm module, as an auxiliary decision-making tool, enhances the robustness of the system in the case of target loss. The YOLOv8 object detection algorithm module is trained using its own dataset and optimizes the weights to enable accurate identification of parachutes. When the OSTrack tracking algorithm module loses the target, the YOLOv8 object detection algorithm module is responsible for re-detecting the parachute and passing the detection results to the OSTrack tracking algorithm module to re-initialize the OSTrack tracking algorithm module.

[0044] In the embodiment of the present invention, the image bit conversion processing algorithm module is used to convert the image bits, including linear or non-linear mapping, normalization processing, and color correction, to ensure that the image details are retained while adapting to the image format requirements of the display device.

[0045] The external delivery device includes an image output device that can be connected through an SDI interface, such as a professional monitor, a display screen with an SDI input interface, etc. The external delivery device is used to receive the data transmitted by the DeckLink video capture card. These data sources are of two types: one is the visible light image data collected by the visible light camera and processed by the industrial computer; the other is the infrared image data collected by the infrared camera and processed successively by the image bit conversion processing algorithm module and the image processing module on the industrial computer. The external delivery device converts the above two types of data into visual images for output.

[0046] In some embodiments, the visible light camera transmits the image data to the DeckLink video capture card through the SDI interface. SDI (Serial Digital Interface) is a digital signal interface widely used for video and audio transmission. The DeckLink video capture card is a hardware device for video capture and output, and it supports the SDI interface. Specifically, the SDI signal output by the visible light camera is transmitted to the SDI input port of the DeckLink video capture card through a supporting connection cable. The DeckLink video capture card receives and decodes the SDI signal transmitted by the visible light camera, converts it into a format recognizable by the computer (such as the YUV or RGB format), and the processed image data is transmitted to the industrial computer. After being processed by the image processing module configured on the industrial computer, it is then transmitted to the DeckLink video capture card and transmitted to the external delivery device through the SDI output port of the DeckLink video capture card. This connection method is used for efficient transmission of image data through the SDI signal and is applicable to visible light cameras.

[0047] In some embodiments, the infrared camera is connected to the industrial control computer via RTSP. RTSP (Real-Time Streaming Protocol) is a protocol for video stream transmission and is widely used for pulling real-time video streams from remote video sources and streaming media servers. Specifically, the image data collected by the infrared camera is transmitted to the industrial control computer via RTSP, processed by the image bit number conversion processing algorithm module and the image processing module configured on the industrial control computer, and then transmitted to the DeckLink video capture card and sent to the external device via the SDI output port of the DeckLink video capture card. This connection method is used to pull image data from the RTSP video source and is applicable to infrared cameras.

[0048] In the embodiments of the present invention, the visible light camera transmits visible light image data to the DeckLink video capture card via SDI signal for processing and then to the industrial control computer, and the image data processed by the industrial control computer is then transmitted to the external device via the DeckLink video capture card. The infrared camera uses the RTSP connection method to pull the infrared image data to the industrial control computer for processing and is sent to the external device via the DeckLink video capture card. The present invention supports both of these connection methods simultaneously to ensure the efficient synchronization and transmission of image data of different types of cameras.

[0049] An intelligent tracking system for a spacecraft parachute based on a vehicle-mounted theodolite provided by the present invention can achieve efficient real-time processing. Using a high-performance industrial control computer and GPU acceleration, a real-time processing speed of over 50fps is achieved. When the frame rate exceeds 50fps, a frame skipping method is adopted, that is, tracking processing is performed every certain number of frames to balance the system load and tracking accuracy, ensure system stability and tracking accuracy, and maintain real-time performance at the same time. At the same time, the present invention can achieve full-automatic tracking and data output. The system realizes full-automatic tracking, and the miss distance (target offset) obtained after tracking can be used for downstream tasks such as spacecraft position correction and data analysis, improving the application value of the system.

[0050] Reference Figures 1 - 3 , on the other hand, the present invention provides an intelligent tracking method for a spacecraft parachute based on a vehicle-mounted theodolite. Based on the above-mentioned intelligent tracking system for a spacecraft parachute based on a vehicle-mounted theodolite, it includes the following steps:

[0051] S1: Image Acquisition: The vehicle-mounted theodolite collects image data of the parachute as the tracking target through a multi-spectral imaging device. Among them, the visible light camera is responsible for capturing clear images of the parachute during the day and under good lighting conditions; the infrared camera provides stable image capture capabilities under low-light or night conditions to capture the infrared image of the parachute. The image data collected by the visible light camera is transmitted to the industrial control computer through the DeckLink video capture card in the SDI interface mode; the image data collected by the infrared camera is transmitted to the industrial control computer through RTSP, ensuring independent processing of image data of different spectra, realizing real-time transmission of image data, and ensuring the synchronization and efficiency of image data.

[0052] S2: Image Preprocessing:

[0053] The image data collected by the visible light camera is directly input into the image processing module configured on the industrial control computer through the connection method of transmitting the SDI signal to the DeckLink video capture card. The visible light image is preliminarily processed, such as denoising and enhancement, and directly used by the OSTrack tracking algorithm module.

[0054] The 14-bit high-bit image data collected by the infrared camera undergoes dynamic range compression and bit-depth conversion through the image bit number conversion processing algorithm module, and is converted into an 8-bit image format. The conversion process includes linear or non-linear mapping, normalization processing, and color correction, ensuring that image details are retained while adapting to the 8-bit image format requirements of the display device.

[0055] S3: Target Tracking: Through the OSTrack tracking algorithm module, the parachute is tracked in real time by combining the shape characteristics of the parachute and the return capsule. The intelligent tracking system continuously monitors the size change of the parachute. If the size of the parachute changes abnormally, that is, the area of the parachute is greater than or equal to 2 times the area before the change, or less than or equal to 0.5 times the area before the change, it is determined that the target is lost and transferred to step S4, which specifically includes the following sub-steps:

[0056] S31: Initialization: The user manually selects the initial position of the parachute through the user interface of the industrial control computer display device, or the intelligent tracking system of the spacecraft parachute based on the vehicle-mounted theodolite automatically detects the position of the parachute and initializes the OSTrack tracking algorithm module;

[0057] S32: Real-time Tracking and Target Loss Detection: The OSTrack tracking algorithm module performs real-time tracking of the parachute based on Vision Transformer and outputs the bounding box coordinates of the tracking target. The intelligent tracking system continuously monitors the size change of the tracking target. If the size of the tracking target changes abnormally, that is, the area of the tracking target is greater than or equal to 2 times the area before the change, or less than or equal to 0.5 times the area before the change, it is determined that the target is lost and transferred to step S4.

[0058] S4: Auxiliary detection: The intelligent tracking system calls the YOLOv8 target detection algorithm module to perform target detection on the current frame image and identify the position of the parachute.

[0059] Furthermore, in step S4, the YOLOv8 target detection algorithm module performs target detection on the current frame image to identify the position of the parachute. It then uses a mistracking correction strategy, combining the shape characteristics of the parachute and the return capsule (an inverted triangle shape. The parachute is large, forming an inverted triangle shape when combined with the return capsule) and their morphological changes under the influence of wind. This strategy ensures correct tracking of the parachute and avoids mistracking of the return capsule through shape matching and feature verification. This strategy includes the following substeps:

[0060] S41: Shape feature extraction: Using image processing technology, through edge detection and contour analysis, the shape features of each target detected by the YOLOv8 target detection algorithm module are extracted to identify whether the target is in the shape of an inverted triangle and whether it meets the characteristics of the parachute and return capsule combination.

[0061] S42: Shape Matching and Verification: The shape features obtained in step S41 are matched with the preset inverted triangle shape features formed by the parachute and return capsule combination, and a shape similarity score is calculated. The shape similarity score is compared with a preset shape similarity score threshold. If the shape similarity score exceeds the threshold, the target is confirmed to be the parachute and return capsule combination. This target position is used as the new tracking target, and the OSTrack tracking algorithm module is reinitialized to continue tracking the target. Otherwise, it is confirmed to be a mistracked target, and the intelligent tracking system ignores the target and continues to call the YOLOv8 object detection algorithm module for redetection until the parachute position is accurately identified.

[0062] Furthermore, according to the changes in the parachute's shape caused by external factors such as wind, the pre-set shape similarity score threshold is dynamically adjusted to ensure that the parachute can still be accurately identified when its shape changes slightly.

[0063] Furthermore, the present invention can realize efficient real-time processing of image data. When the frame rate of the image data exceeds 50fps, a frame extraction method is adopted, that is, tracking processing is performed every set number of frames (for example, 5 frames, 8 frames or 10 frames) to ensure the stability and tracking accuracy of the intelligent tracking system.

[0064] Furthermore, the present invention stores the tracking data (including the target's bounding box coordinates, miss distance, etc.) in a storage module (such as a solid-state drive) for use by downstream tasks.

[0065] The present invention can display the tracking results in real time, and users can view the real-time position of the parachute through a display device; it allows users to manually select the tracking target, adjust the tracking parameters or reset the tracking system; it provides a real-time monitoring function, and users can view the tracking status and the performance of the tracking system at any time.

[0066] The present invention supports the function of pulling RTSP protocol images, meets the access requirements of multi-source video streams, and has the function of converting 14-bit infrared images to 8-bit display, improving the image processing and display effects. The present invention uses the OSTrack tracking algorithm as the main tracking algorithm and introduces the YOLOv8 object detection algorithm as an auxiliary decision-making tool. When the tracking target is lost, the YOLOv8 object detection algorithm can be called to re-detect and identify the parachute in time, ensuring the continuity, accuracy and real-time performance of the tracking. The present invention has high-efficiency real-time processing capabilities. By using a high-performance industrial computer and GPU acceleration, a real-time processing speed exceeding 50fps is achieved, meeting the requirements of high-frame-rate tasks, and the stability and tracking accuracy of the tracking system are ensured through the frame extraction method. In addition, the present invention combines an anti-mis-tracking correction strategy. Through shape matching and feature verification, it ensures that the tracking system can accurately distinguish the parachute from the return capsule, avoiding mis-tracking the return capsule under the influence of wind, and further improving the robustness and safety of the tracking system in complex environments. The system of the present invention has a high degree of integration and supports two connection methods at the same time, ensuring the efficient synchronization and transmission of camera image data of different types, realizing the efficient transmission and synchronous processing of visible light image data and infrared image data, and reducing the complexity and maintenance cost of the system. The tracking system of the present invention can achieve full-automatic tracking, and the tracking data obtained after tracking (including the bounding box coordinates of the target, the amount of off-target, etc.) can be effectively used for downstream tasks, further improving the application value and engineering feasibility of the tracking system. The present invention balances the real-time performance and accuracy of the tracking system. Through the early candidate elimination module and efficient algorithm design, while ensuring the tracking accuracy, the real-time processing capabilities of the tracking system are improved to adapt to various application scenarios. The present invention solves the problems of the existing parachute tracking system in terms of light dependence, target mis-identification, environmental adaptability, real-time processing capabilities and feature learning, and provides an intelligent tracking system and tracking method with high precision, strong robustness and high real-time processing capabilities, significantly improving the safety and reliability during the spacecraft landing process.

[0067] It should be understood that the various forms of the processes shown above can be used, steps can be reordered, added or deleted. For example, the steps recorded in the disclosure of the present invention can be executed in parallel, sequentially or in a different order, as long as the results expected by the technical solution disclosed in the present invention can be achieved. No limitation is made herein.

[0068] The above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent tracking system for spacecraft parachutes based on vehicle-mounted theodolites, characterized in that, It includes a vehicle-mounted theodolite, an external delivery device, an industrial control computer, a video capture card, an image processing module, and an image bit conversion processing algorithm module for converting the image bits. The vehicle-mounted theodolite is connected to the industrial control computer, and the video capture card is connected to the industrial control computer, the vehicle-mounted theodolite, and the external delivery device. The image processing module and the image bit conversion processing algorithm module are configured on the industrial control computer, and the image processing module and the image bit conversion processing algorithm module can be run through the industrial control computer. Among them, the vehicle-mounted theodolite includes a theodolite body and a multispectral imaging device, and the multispectral imaging device is integrally installed on the theodolite body to collect the image data and position data of the parachute in real time. The industrial control computer includes a processor, a memory, a graphics processor, a display device, and a storage module to perform real-time processing of the image data, display the tracking result in real time, and store the data. The image processing module includes an OSTrack tracking algorithm module and a YOLOv8 object detection algorithm module to perform real-time tracking and auxiliary detection of the parachute. Through the OSTrack tracking algorithm module, the parachute is tracked in real time by combining the shape characteristics of the parachute and the spacecraft. The intelligent tracking system continuously monitors the size change of the parachute. If the size of the parachute changes abnormally, that is, the area of the parachute is greater than or equal to 2 times the area before the change, or less than or equal to 0.5 times the area before the change, it is determined that the target is lost and transferred to auxiliary detection. Auxiliary detection: The intelligent tracking system calls the YOLOv8 object detection algorithm module to perform object detection on the current frame image, identify the position of the parachute, use this position as a new tracking target, and re-initialize the OSTrack tracking algorithm module to continue tracking; otherwise, it is confirmed as a mis-tracked target, and the intelligent tracking system ignores this target and continues to call the YOLOv8 object detection algorithm module for re-detection until the position of the parachute is accurately identified. The auxiliary detection includes: When using the YOLOv8 object detection algorithm module to perform object detection on the current frame image to identify the position of the parachute, an anti-mis-tracking correction strategy is adopted, and by combining the inverted triangle shape characteristics of the parachute and the spacecraft and the morphological changes under the influence of wind, through shape matching and feature verification, the correct tracking of the parachute is ensured and the return capsule is prevented from being mis-tracked. It specifically includes the following sub-steps: S41: Shape feature extraction: Using image processing technology, through edge detection and contour analysis, extract the shape features of each target detected by the YOLOv8 object detection algorithm module. S42: Shape matching and verification: Match the shape features obtained in step S41 with the inverted triangle shape features formed by the preset combination of the parachute and the spacecraft, and calculate the shape similarity score; Compare the shape similarity score with the preset shape similarity score threshold. If the shape similarity score is higher than the threshold, confirm that the target is the combination of the parachute and the return capsule, use this target position as the new tracking target, re-initialize the OSTrack tracking algorithm module, and achieve continuous tracking; Otherwise, confirm it as a mis-tracked target, and the intelligent tracking system ignores this target and continues to call the YOLOv8 target detection algorithm module for re-detection until the position of the parachute is accurately identified; Among them, according to the morphological changes of the parachute, the preset shape similarity score threshold is dynamically adjusted.

2. The intelligent tracking system for spacecraft parachute based on vehicle-mounted theodolite according to claim 1, wherein When the OSTrack tracking algorithm module loses the target, the YOLOv8 target detection algorithm module serves as an auxiliary decision-making tool, responsible for re-detecting the parachute, and transmitting the detection result to the OSTrack tracking algorithm module to re-initialize the OSTrack tracking algorithm module.

3. The intelligent tracking system for spacecraft parachute based on vehicle theodolite according to claim 2, characterized in that, The OSTrack tracking algorithm module uses the single-stream object tracking algorithm OSTrack based on ViT as the tracking algorithm, and has the following optimizations and extensions: Joint feature learning and relationship modeling, by splicing the template image and the search area image and inputting them into the ViT encoder, using the self-attention mechanism to distinguish the target from the background during the encoding stage, and effectively generating discriminative target features; Early candidate elimination module, insert an early candidate elimination module in the multi-layer encoder of ViT, and use the similarity score between the template image and the search area image to dynamically eliminate background area candidates; Mis-track correction, combined with the inverted triangle shape features of the parachute and the spacecraft and the morphological changes under the influence of wind, through shape matching and feature verification strategies, ensure correct tracking of the parachute and avoid mis-tracking the spacecraft.

4. The intelligent tracking system for spacecraft parachute based on vehicle-mounted theodolite according to claim 2, wherein, The YOLOv8 target detection algorithm module is trained using its own dataset and optimizes the weights to accurately identify the parachute.

5. The intelligent tracking system for spacecraft parachutes based on vehicle-mounted theodolites according to claim 1, characterized in that, The multi-spectral imaging device includes a visible light camera and an infrared camera; The visible light camera transmits the image data to the video capture card through the SDI interface method. After the video capture card receives and processes the visible light image data, it is transmitted to the industrial control computer. After being processed by the image processing module configured on the industrial control computer, it is then transmitted to the external device through the video capture card; The infrared camera is connected to the industrial control computer through RTSP. The image data collected by the infrared camera is transmitted to the industrial control computer through RTSP. After being processed by the image bit conversion processing algorithm module and the image processing module configured on the industrial control computer, it is then transmitted to the external device through the video capture card.

6. An intelligent tracking method for a spacecraft parachute based on a vehicle-mounted theodolite, characterized in that, The vehicle-mounted theodolite-based spacecraft parachute intelligent tracking system according to any one of claims 1-5, comprising: S1: Image acquisition: The vehicle-mounted theodolite acquires the image data of the parachute through the multi-spectral imaging device. The visible light image data acquired by the multi-spectral imaging device is transmitted to the industrial control computer through the video capture card using the SDI interface method, and the infrared image data acquired by the multi-spectral imaging device is transmitted to the industrial control computer through RTSP; S2: Image preprocessing: The visible light image data is directly input into the image processing module configured on the industrial control computer through the connection method of transmitting via the SDI signal to the video capture card. The visible light image is preliminarily processed and directly used by the OSTrack tracking algorithm module; the 14-bit high-bit image data collected by the infrared camera is processed through the image bit conversion processing algorithm module and converted into an 8-bit image format; the frame rate of the image data is identified. When the frame rate of the image data exceeds 50 fps, the frame extraction method is adopted; S3: Target tracking: Through the OSTrack tracking algorithm module, the parachute is tracked in real time by combining the shape characteristics of the parachute and the spacecraft; the intelligent tracking system continuously monitors the size change of the parachute. If the size of the parachute changes abnormally, that is, the area of the parachute is greater than or equal to 2 times the area before the change, or less than or equal to 0.5 times the area before the change, it is determined that the target is lost, and the process proceeds to step S4; S4: Auxiliary detection: The intelligent tracking system calls the YOLOv8 target detection algorithm module to perform target detection on the current frame image, identify the position of the parachute, use this position as the new tracking target, and re-initialize the OSTrack tracking algorithm module to achieve continuous tracking; Step S4 includes: When the YOLOv8 target detection algorithm module is used to perform target detection on the current frame image to identify the position of the parachute, an anti-mis-tracking correction strategy is adopted. Combining the inverted triangle shape characteristics of the parachute and the spacecraft and the morphological changes under the influence of wind, through shape matching and feature verification, ensure correct tracking of the parachute and avoid mis-tracking the return capsule. Specifically, it includes the following sub-steps: S41: Shape feature extraction: Using image processing technology, through edge detection and contour analysis, extract the shape features of each target detected by the YOLOv8 target detection algorithm module; S42: Shape matching and verification: Match the shape features obtained in step S41 with the inverted triangle shape features formed by the preset combination of the parachute and the spacecraft, and calculate the shape similarity score; Compare the shape similarity score with the preset shape similarity score threshold. If the shape similarity score is higher than the threshold, confirm that the target is a combination of the parachute and the return capsule, use this target position as the new tracking target, and re-initialize the OSTrack tracking algorithm module to achieve continuous tracking; otherwise, confirm it as a mis-tracking target, and the intelligent tracking system ignores this target and continues to call the YOLOv8 target detection algorithm module for re-detection until the position of the parachute is accurately identified; Among them, according to the morphological changes of the parachute, the preset shape similarity score threshold is dynamically adjusted.

7. The intelligent tracking method for a spacecraft parachute based on a vehicle-mounted theodolite according to claim 6, wherein In step S1, the visible light camera of the multispectral imaging device is responsible for capturing clear images of the parachute during the day and under good lighting conditions, while the infrared camera of the multispectral imaging device provides stable image capture capabilities under low light or night conditions to capture infrared images of the parachute.

8. The intelligent tracking method for the spacecraft parachute based on vehicle-mounted theodolite according to claim 6, wherein, Step S3 includes the following steps: S31: Initialization: Manually select the initial position of the parachute through the user interface of the industrial control computer display device, or automatically detect the parachute position by the intelligent tracking system of the spacecraft parachute based on the vehicle-mounted theodolite, and initialize the OSTrack tracking algorithm module; S32: Real-time tracking and target loss detection: Real-time track the parachute through the OSTrack tracking algorithm module and output the bounding box coordinates of the parachute; The intelligent tracking system continuously monitors the size change of the parachute. If the size of the parachute changes abnormally, that is, the area of the parachute is greater than or equal to 2 times the area before the change, or less than or equal to 0.5 times the area before the change, it is determined that the target is lost, and go to step S4.

9. The intelligent tracking method for the spacecraft parachute based on the vehicle-mounted theodolite according to claim 6, wherein, The intelligent tracking system stores the tracking data during the tracking process into the storage module of the industrial control computer; The user can view the real-time position of the parachute through the display device; The user can manually select the tracking target, adjust or reset the tracking system, and view the tracking status and tracking system performance at any time.

Images