Target ranging system based on multi-angle camera collaborative perception

Abstract

The invention provides a target ranging system based on multi-angle camera collaborative perception, and the system comprises a wide-angle camera module which is used for collecting a large-view-field scene image and obtaining an initial pixel coordinate of a target in a wide-angle camera pixel coordinate system; the rotatable high-resolution camera module has a function of dynamically adjusting a pitch angle beta and a yaw angle alpha around an optical axis, and is used for adjusting a visual angle according to the initial pixel coordinates so as to track a target; the calculation control module is used for performing distortion correction on the images of the wide-angle camera and the high-resolution camera; the rotation angle of the high-resolution camera is controlled, and the visual angle is dynamically adjusted to track the target; based on the normalized coordinates of the wide-angle camera and the multi-view geometric constraint of the high-resolution camera, a nonlinear equation set is constructed, and the absolute distance from the target to the wide-angle camera is solved through a numerical analysis method.

Description

Technical Field

[0001] The present invention belongs to the technical fields such as the application of computer vision in dynamic environment perception and the design of multi-camera collaborative ranging systems, and specifically relates to a target ranging system based on multi-angle camera collaborative perception. Background Art

[0002] Existing binocular ranging technologies are widely used in fields such as computer vision, robotics, and autonomous driving. They mainly rely on methods such as parallax calculation, feature matching, or deep learning to estimate the target distance. However, traditional binocular systems require the camera optical axes to be fixedly parallel, with a rigid structure and difficulty in adapting to dynamic environments. In addition, due to the fixed baseline, the ranging accuracy decreases as the distance increases, resulting in significant ranging errors for distant targets. Moreover, binocular ranging highly depends on image feature matching. In cases of weak texture, occlusion, or lighting changes, it is difficult to guarantee the matching accuracy, thereby affecting the accuracy of depth estimation. Even when using deep learning methods, there are problems of large computational volume and high requirements for hardware, making it difficult to run in real time on embedded devices or low-power platforms.

[0003] On the other hand, existing binocular ranging methods have poor adaptability in dynamic scenarios and lack the ability to actively adjust the camera angles, resulting in the inability to continue ranging after the target moves out of the field of view. In addition, the ranging range is limited by the field of view of the binocular system, unable to flexibly cover a large range of targets and prone to ranging blind spots. The computational complexity of traditional methods is relatively high and the real-time performance is insufficient, making it difficult to apply especially on mobile platforms or resource-constrained devices. Therefore, there is an urgent need for a more flexible, wider-ranging, and complex environment-adaptable ranging solution to overcome the limitations of existing technologies and improve the robustness and practicality of the ranging system. Summary of the Invention

[0004] In view of the defects and deficiencies existing in the prior art, the present invention provides a target ranging system and method based on multi-angle camera collaborative perception, and its innovative design points include:

[0005] Dynamic collaborative ranging architecture: By the rough positioning of the large field of view of the wide-angle camera and the dynamic perspective adjustment (pitch angle β, yaw angle α) of the rotatable high-resolution camera, the limitation of the fixed baseline of the traditional binocular system is broken through, and blind-spot-free and high-precision target tracking and ranging are realized;

[0006] Multi-view geometric modeling and simultaneous solution: Based on the normalized coordinates (X p / Z p , Y p / Z p ) of the wide-angle camera and the dynamic geometric constraints (spatial distance O2B, projection plane parameter PB, angular relationship ∠PBQ2) of the high-resolution camera, a non-linear equation system is constructed and the absolute distance Z is solved by numerical analysis methods (such as Newton iteration method)p , significantly improving the ranging accuracy;

[0007] Embedded software and hardware co - optimization: Integrate parallel computing for image distortion correction, real - time feedback perspective control, and lightweight ranging calculation modules, adapt to low - power hardware platforms, and support real - time processing;

[0008] Enhanced robustness in complex environments: Reduce the dependence on feature matching through multi - perspective data fusion (redundant constraints at least at two rotation angles), and maintain ranging stability in weak - texture, occluded, and dynamic scenes;

[0009] Full - scene applicability: Cover dynamic requirements such as obstacle tracking in autonomous driving, target positioning in security monitoring, and UAV mapping, and achieve large - range environmental perception and high - precision ranging capabilities.

[0010] The present invention takes dynamic perspective adjustment and multi - perspective geometric modeling as the core, and systematically solves the bottleneck problems of traditional binocular ranging, such as poor flexibility, low accuracy, and complex calculation, through hardware cooperation, algorithm optimization, and scene adaptation.

[0011] The technical solution specifically adopted by the present invention to solve its technical problems is:

[0012] A target ranging system based on multi - angle camera collaborative perception, comprising:

[0013] A wide - angle camera module, used to collect large - field - of - view scene images and obtain the initial pixel coordinates of the target in the pixel coordinate system of the wide - angle camera;

[0014] A rotatable high - resolution camera module, having the function of dynamically adjusting the pitch angle β and yaw angle α around the optical axis, and used to adjust the perspective to track the target according to the initial pixel coordinates;

[0015] A calculation and control module, used for:

[0016] Perform distortion correction on the images of the wide - angle camera and the high - resolution camera;

[0017] Control the rotation angle of the high - resolution camera and dynamically adjust the perspective to track the target;

[0018] Based on the normalized coordinates of the wide - angle camera and the multi - perspective geometric constraints of the high - resolution camera, construct a non - linear equation system and solve the absolute distance from the target to the wide - angle camera through numerical analysis methods.

[0019] Furthermore, constructing the non - linear equation system based on the normalized coordinates of the wide - angle camera and the multi - perspective geometric constraints of the high - resolution camera specifically includes:

[0020] The direction of the optical axis is adjusted according to the pitch angle β and the yaw angle α of the high-resolution camera, and the geometric relationship of the target point on the projection plane of the optical axis after rotation (NPM plane) is established;

[0021] Combined with the spatial distance between the target point and the optical center of the high-resolution camera The vertical distance between the target point and the optical axis projection plane And the angle between the target point and the optical axis Construct dynamic geometric constraints;

[0022] The target normalized coordinates provided by the wide-angle camera are combined with the dynamic geometric constraints to form a nonlinear equation group.

[0023] Furthermore, the numerical analysis method is Newton's iteration method, which is used to solve the nonlinear equations:

[0024] According to the rotation angle (α, β) of the high-resolution camera and the pixel coordinates (x 2p ,y 2p ), calculate the dynamic geometric parameters (BM) on the optical axis projection plane;

[0025] The normalized coordinate (X p / Z p ,Y p / Z p ) mark is combined with the dynamic geometric parameters to form a parameterized nonlinear equation;

[0026] The convergence of the equation is calculated by Newton iteration method, and the absolute distance Z from the target to the wide-angle camera is output. p .

[0027] Furthermore, the wide-angle camera and the high-resolution camera are initially parallel to each other in optical axis, and their optical centers are collinear. The baseline distance b is a fixed parameter. When calculating the spatial coordinates of the target point P, the baseline distance b is directly introduced into the geometric constraint equation.

[0028] Furthermore, the computing control module integrates an embedded optimization algorithm, including:

[0029] Parallel computing of image distortion correction;

[0030] Real-time feedback control of viewing angle adjustment;

[0031] Lightweight processing of ranging solution to adapt to low-power hardware platforms.

[0032] Furthermore, the calculation control module improves the ranging robustness by:

[0033] The direction information of the target in the pixel coordinate system is obtained through a wide-angle camera;

[0034] Controlling a high-resolution camera to capture target images at at least two different rotation angles to obtain corresponding pixel coordinates;

[0035] The geometric constraint equations under different rotation angles are combined to generate redundant spatial constraints to solve the absolute distance.

[0036] Furthermore, when the target is within the field of view of the wide-angle camera, the high-resolution camera adjusts the pitch angle β and the yaw angle α based on the initial pixel coordinates of the target provided by the wide-angle camera, so that the target enters the field of view of the high-resolution camera, and solves the absolute distance through the multi-view geometric constraint equation; wherein the adjustment angle of the high-resolution camera includes at least one combination of the pitch angle β and the yaw angle α, so that the target is within the field of view of the high-resolution camera, and different rotation angle combinations correspond to different multi-view geometric constraints, and ranging can be achieved.

[0037] The above systems are suitable for dynamic target ranging scenarios, including autonomous driving obstacle tracking, security monitoring target positioning, and drone long-distance mapping.

[0038] And, a target ranging method based on multi-angle camera collaborative perception, comprising the following steps:

[0039] Wide-angle image acquisition and correction: Use a wide-angle camera to acquire large-field-of-view scene images, perform distortion correction, and obtain the initial pixel coordinates of the target in the wide-angle camera pixel coordinate system;

[0040] Dynamic adjustment of the high-resolution camera: according to the initial pixel coordinates, the rotatable high-resolution camera is controlled to adjust the pitch angle β and the yaw angle α around the optical axis so that the target enters its field of view;

[0041] Multi-view image acquisition and modeling: Capture high-resolution images at different rotation angles and perform distortion correction to obtain the pixel coordinates of the target in the pixel coordinate system of the high-resolution camera;

[0042] Multi-view geometric constraint construction: Based on the normalized coordinates of the wide-angle camera and the dynamic geometric relationship of the high-resolution camera, a nonlinear equation group including the pitch angle β and the yaw angle α is established;

[0043] Absolute distance solution: Solve the nonlinear equations by numerical analysis method and output the absolute distance Z from the target to the wide-angle camera. p .

[0044] Furthermore, the construction of the multi-view geometric constraints includes:

[0045] According to the pitch angle β and the yaw angle α, the dynamic projection relationship between the target point and the optical axis of the high-resolution camera is established;

[0046] Combine the baseline distance b and the rotation angle parameter to generate the spatial distance and angle constraint equation;

[0047] The numerical analysis method is the Newton iteration method, and the convergent solution of the nonlinear equation is realized through parametric matching.

[0048] Further, when the target is within the field of view of the wide-angle camera, the steps of dynamically adjusting the high-resolution camera to the absolute distance calculation are repeatedly executed, and the ranging result is updated through the multi-view geometric constraints of the high-resolution camera at different rotation angles.

[0049] Compared with the prior art, the beneficial effects of the present invention and its preferred solutions at least include:

[0050] Significantly improve the ranging flexibility and scene adaptability: By dynamically adjusting the viewing angle (pitch angle, yaw angle) of the rotatable high-resolution camera, break through the limitations of the fixed baseline and fixed field of view of the traditional binocular system, effectively eliminate the ranging blind area, and meet the requirements of dynamic target tracking and large-scale scene coverage;

[0051] Enhance the long-distance ranging accuracy: Based on the multi-view geometric constraint modeling and the simultaneous solution of nonlinear equations, combine the direction information of the wide-angle camera with the dynamic space constraints of the high-resolution camera, and reduce the long-distance error accumulation caused by the fixed baseline in the traditional parallax method;

[0052] Optimize the calculation efficiency and real-time performance: Through the collaborative design of embedded software and hardware, integrate image distortion correction, viewing angle adjustment and lightweight calculation process, adapt to the low-power hardware platform, and achieve real-time high-precision ranging in complex scenarios;

[0053] Improve the robustness in complex environments: Use multi-view data fusion (such as redundant space constraints) to reduce the dependence on single-image feature matching, and maintain the ranging stability in weak texture, occlusion or light change scenarios;

[0054] Expand the application scenario coverage ability: Support the continuous tracking and high-precision positioning of targets in dynamic scenarios such as autonomous driving, UAV navigation, and security monitoring, and balance the requirements of large field of view perception and fine ranging;

[0055] This solution solves the rigid defects of traditional systems with a dynamic collaborative architecture, balances accuracy and efficiency with multi-view modeling and optimized calculation, and finally realizes the ranging performance upgrade of "flexible-precise-efficient-reliable". BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The present invention will be further described in detail below in conjunction with the drawings and specific embodiments:

[0057] Figure 1 It is a system diagram of target ranging for the embodiment of the present invention facing multi-angle camera collaborative perception;

[0058] Figure 2Schematic diagram of the situation where the target object P in the embodiment of the present invention is at the upper left of the wide-angle large field-of-view camera;

[0059] Figure 3 Schematic diagram of the situation where the target object P in the embodiment of the present invention is at the upper right of the wide-angle large field-of-view camera;

[0060] Figure 4 Schematic diagram of the situation where the target object P in the embodiment of the present invention is at the lower right of the wide-angle large field-of-view camera. Detailed implementation manners

[0061] To make the features and advantages of this patent more obvious and understandable, specific embodiments are hereinafter given and described in detail as follows:

[0062] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs.

[0063] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary implementation manners according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0064] The present invention has systematically optimized and improved the deficiencies of traditional binocular ranging technology in terms of flexibility, ranging range, feature matching dependence, computational efficiency, and dynamic adaptability. Traditional binocular ranging systems usually require the optical axes of the two cameras to be fixedly parallel, which not only results in a strong rigid structure of the system and makes it difficult to adapt to complex dynamic environments, but also limits the ranging range by the fixed baseline length. Especially in long-distance ranging, the error gradually increases, and even effective depth information cannot be obtained, resulting in the ranging blind area problem. In addition, traditional binocular ranging highly depends on image feature matching. In an environment with weak texture, low contrast, occlusion, or drastic illumination changes, the accuracy of feature point matching drops significantly, resulting in unstable or even invalid depth estimation. On the other hand, traditional methods usually adopt ranging algorithms based on disparity calculation, which have a high computational complexity. Especially when processing high-resolution images, they require a large amount of computational resources and storage space and are difficult to achieve real-time operation on embedded devices or low-power platforms. In addition, traditional binocular systems cannot dynamically adjust the viewing angle. When the target exceeds the overlapping field of view of the binocular cameras, the system cannot continuously track and range, and it is difficult to adapt to the ranging errors caused by target movement or the movement of the cameras themselves.

[0065] To solve the above problems, the present invention adopts a cooperative ranging scheme of a wide-angle camera + a rotatable high-resolution camera, fully combining the advantages of the two cameras to achieve high-precision ranging. The wide-angle camera is responsible for providing a large-range scene perception and quickly locking the approximate position of the target, while the high-resolution camera can dynamically adjust the viewing angle (pitch and yaw rotation) to keep the target always within the field of view, thereby obtaining high-precision depth information. This cooperative ranging method effectively solves the ranging blind area problem and still maintains high precision when ranging long-distance targets. At the same time, this system reduces the dependence on single feature matching and improves the ranging robustness through multi-view data fusion, enabling it to still operate stably in complex environments (such as low light, sparse texture, dynamic targets). In addition, the calculation control module of the present invention optimizes the image correction, viewing angle adjustment, and ranging calculation processes, improving the overall calculation efficiency and ensuring that the system can operate efficiently on embedded devices or low-power platforms, providing a flexible, efficient, and highly adaptable solution for high-precision ranging.

[0066] The target ranging system based on multi-angle camera cooperative perception in the embodiment of the present invention is as Figure 1 shown, mainly including an image acquisition module, a calculation control module, and a multi-view ranging method to improve the flexibility and precision of the ranging system.

[0067] Image acquisition module: It is composed of a wide-angle large field-of-view camera and a high-resolution camera respectively. The wide-angle large field-of-view camera is used to collect a wider field of view for target exploration and target ranging, and the high-resolution camera collects high-resolution images. This camera can rotate arbitrarily in pitch and yaw around the optical axis.

[0068] Calculation control module: Perform image distortion correction on the images collected by the wide-angle large field-of-view camera to obtain the image pixel coordinates of the target object in the wide-angle large field-of-view camera; then control the high-resolution camera to rotate at any angle. As long as the target object appears within the field of view, receive the images collected by the high-resolution camera, perform distortion correction on the images, and obtain the pixel coordinates of the target object appearing in the high-resolution camera; then the absolute distance of the target object from the wide-angle large field-of-view camera can be calculated according to the image pixel coordinates of the target object in the two cameras after image correction.

[0069] The detailed working process of the system is as follows:

[0070] 1. First, the image acquisition module collects the image of the target object appearing in the wide-angle large field-of-view camera, performs image distortion correction, and obtains the pixel coordinates of the target object in the image pixel coordinate system of this camera.

[0071] 2. The calculation control module controls the high-resolution camera to rotate at any pitch angle and yaw angle.

[0072] 3. The image acquisition module captures images of the target object appearing within the field of view of the high-resolution camera at different rotation angles and corrects image distortion; it obtains the pixel coordinates of the target object in the pixel coordinate system of the high-resolution camera images at different rotation angles.

[0073] 4. The calculation and control module calculates the absolute distance from the target object to the wide-angle large field-of-view camera based on the target pixel coordinates obtained by the wide-angle large field-of-view camera and the target pixel coordinates obtained by the high-resolution camera at different rotation angles.

[0074] The following further specifically describes the target ranging method based on multi-angle camera collaborative perception used in this system.

[0075] As Figure 2 shown, in this embodiment, the optical axes of the wide-angle large field-of-view camera and the high-resolution camera are parallel and the optical centers are on the same straight line in the initial state. The optical center positions of the two cameras are O1 and O2 respectively; the focal lengths of the two cameras are f1 and f2 respectively; the baseline distance between the two cameras is b; among them, the wide-angle large field-of-view camera is fixed. The coordinates of the wide-angle large field-of-view camera in its photosensitive plane pixel coordinate system u1-v1 are denoted as (C u1 , C v1 ); the coordinates of the target point in the photosensitive plane pixel coordinate system u1-v1 of the wide-angle large field-of-view camera are denoted as (x 1p , y 1p ); the pixel size (the actual distance corresponding to each pixel) of the wide-angle large field-of-view camera is denoted as d1; the horizontal yaw angle of the high-resolution camera around its optical center is denoted as α and positive to the left, and the vertical rotation angle around its optical center is denoted as β and positive upward; the coordinates of the high-resolution camera in its photosensitive plane pixel coordinate system u2-v2 are denoted as (C u2 , C v2 ), and the coordinates of the target point in the photosensitive plane pixel coordinate system u2-v2 of the wide-angle large field-of-view camera are denoted as (x 2p , y 2p ); the pixel size (the actual distance corresponding to each pixel) of the high-resolution camera is denoted as d2; the origin of the world coordinate system X-Y-Z coincides with the camera coordinate system on the optical center O1 of the wide-angle large field-of-view camera. The positive direction of the Z-axis is along the optical axis direction passing through the origin, the positive direction of the Y-axis is vertically upward, and the positive direction of the X-axis is perpendicular to the optical axis direction and to the left.

[0076] Let the coordinates of the target object P in the world coordinate system be (X p , Y p , Z p ). According to the imaging formula of the target point P in the large field-of-view camera, we have:

[0077]

[0078]

[0079] As Figure 2 shown, project the points on the photosensitive plane of the high-resolution camera through the optical center onto the NPM plane, where NP⊥NM. Point N is the intersection of the extension line of the connection line of the points on the axis projected by the target object P on the high-resolution camera through the optical center with the NPM plane, point M is the intersection of the optical axis of the high-resolution camera rotating at different pitch and yaw angles with the plane NPM, and point B is the intersection of the vertical plane where the target object P is located and the optical axis of the high-resolution camera.

[0080]

[0081] By combining equations (1), (2), and (3), two equation equations for the absolute distance Z can be obtained:

[0082]

[0083] Applying the parameter matching method in numerical analysis, the absolute distance Z between the wide-angle large field-of-view camera and the target object can be solved at different rotation angles of the high-resolution camera.

[0084] As Figure 3 、 4 shown, when the target object P is located in the upper right or lower right of the wide-angle camera, equation (4) can still be obtained from the imaging relationship of the target object P on the high-resolution camera, proving that the derived equation is applicable to measuring the absolute distance Z of the target object at any position of the wide-angle camera.

[0085] The device product obtained through the solution of this embodiment is a multi-view ranging system based on the cooperation of a wide-angle camera and a high-resolution camera. It combines the advantages of the large field of view of the wide-angle camera and the fine ranging ability of the high-resolution camera, and has the advantages of high flexibility, wide ranging range, and adaptability to complex environments. Compared with the traditional fixed binocular ranging system, this solution allows the high-resolution camera to rotate freely without strictly aligning with the target, greatly improving the target capture ability and reducing the ranging blind area. In addition, the system can accurately measure the distance of all targets within the field of view of the wide-angle camera. Even if the target is at a long distance or in a complex background environment, high-precision depth information can still be obtained. The calculation and control module can correct the image distortion in real time and improve the ranging accuracy through the multi-view ranging method, enabling the system to adapt to the ranging requirements of dynamic targets, while reducing the dependence on feature matching and improving the robustness.

[0086] This solution is applicable to a variety of scenarios that require precise distance measurement, such as autonomous driving, robot navigation, security monitoring, maritime search and rescue, forest fire warning and other fields. In terms of intelligent monitoring, it can be used for large-scale deployment and automatically adjust high-resolution cameras for precise distance measurement when detecting suspicious targets; in autonomous driving and robot applications, it can provide more reliable environmental perception and enhance obstacle avoidance and path planning capabilities; in the field of emergency rescue, it can measure the target position at a long distance and provide important information for rescue decision-making. In addition, this solution can also be used in professional application scenarios such as surveying and mapping, UAV distance measurement, etc., providing efficient and precise distance measurement solutions for a wide range of industries.

[0087] It should be noted that unless otherwise defined, the technical terms or scientific terms used in this invention should have the ordinary meanings understood by those with ordinary skills in the field to which this invention belongs. The "first", "second" and similar terms used in this invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. Words such as "connected" or "linked" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right", etc. are only used to represent relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0088] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still fall within the protection scope of the technical solution of the present invention.

[0089] This patent is not limited to the above best implementation manner. Anyone can derive various other forms of target distance measurement systems based on multi-angle camera collaborative perception under the inspiration of this patent. All equal changes and modifications made according to the scope of the patent application of the present invention shall fall within the coverage of this patent.

Claims

1. A target ranging system based on multi-angle camera collaborative perception, characterized in that: include: A wide-angle camera module is used to capture a scene image with a large field of view and obtain the initial pixel coordinates of the target in the pixel coordinate system of the wide-angle camera; A rotatable high-resolution camera module having a dynamic adjustment function of a pitch angle β and a yaw angle α around an optical axis, for adjusting the viewing angle according to the initial pixel coordinates to track the target; Computing control module for: Distortion correction for images from wide-angle and high-resolution cameras; Control the rotation angle of the high-resolution camera and dynamically adjust the viewing angle to track the target; Based on the normalized coordinates of the wide-angle camera and the multi-view geometric constraints of the high-resolution camera, a nonlinear equation group is constructed and the absolute distance from the target to the wide-angle camera is solved by numerical analysis method.

2. The target ranging system based on multi-angle camera collaborative perception according to claim 1 is characterized in that: The construction of the nonlinear equations based on the normalized coordinates of the wide-angle camera and the multi-view geometric constraints of the high-resolution camera specifically includes: The direction of the optical axis is adjusted according to the pitch angle β and the yaw angle α of the high-resolution camera, and the geometric relationship of the target point on the projection plane of the optical axis after rotation is established; Dynamic geometric constraints are constructed by combining the spatial distance between the target point and the optical center of the high-resolution camera, the vertical distance between the target point and the optical axis projection plane, and the angle between the target point and the optical axis. The target normalized coordinates provided by the wide-angle camera are combined with the dynamic geometric constraints to form a nonlinear equation group.

3. The target ranging system based on multi-angle camera collaborative perception according to claim 2 is characterized in that: The numerical analysis method is Newton's iteration method, which is used to solve the nonlinear equations: According to the rotation angle of the high-resolution camera and the pixel coordinates of the target point, the dynamic geometric parameters on the optical axis projection plane are calculated; combining the normalized coordinates of the wide-angle camera with the dynamic geometric parameters to form a parameterized nonlinear equation; The convergence of the equation is calculated by Newton iteration method, and the absolute distance Z from the target to the wide-angle camera is output. p .

4. The target ranging system based on multi-angle camera collaborative perception according to claim 1, characterized in that: The wide-angle camera and the high-resolution camera have initial optical axes parallel to each other, and their optical centers are collinear. The baseline distance b is a fixed parameter. When calculating the spatial coordinates of the target point P, the baseline distance b is directly introduced into the geometric constraint equation.

5. The target ranging system based on multi-angle camera collaborative perception according to claim 1, characterized in that: The computing control module integrates an embedded optimization algorithm, including: Parallel computing of image distortion correction; Real-time feedback control of viewing angle adjustment; Lightweight processing of ranging solution to adapt to low-power hardware platforms.

6. The target ranging system based on multi-angle camera collaborative perception according to claim 1, characterized in that: The calculation control module improves the ranging robustness by: The direction information of the target in the pixel coordinate system is obtained through a wide-angle camera; Controlling a high-resolution camera to capture target images at at least two different rotation angles to obtain corresponding pixel coordinates; The geometric constraint equations under different rotation angles are combined to generate redundant spatial constraints to solve the absolute distance.

7. The target ranging system based on multi-angle camera collaborative perception according to claim 1, characterized in that: When the target is within the field of view of the wide-angle camera, the high-resolution camera adjusts the pitch angle β and the yaw angle α based on the initial pixel coordinates of the target provided by the wide-angle camera, so that the target enters the field of view of the high-resolution camera, and solves the absolute distance through the multi-view geometric constraint equation; wherein the adjustment angle of the high-resolution camera includes at least one combination of the pitch angle β and the yaw angle α, so that the target is within the field of view of the high-resolution camera, and different rotation angle combinations correspond to different multi-view geometric constraints, and ranging can be achieved.

8. A target ranging method based on multi-angle camera collaborative perception, characterized in that: The following steps are involved: Wide-angle image acquisition and correction: Use a wide-angle camera to acquire large-field-of-view scene images, perform distortion correction, and obtain the initial pixel coordinates of the target in the wide-angle camera pixel coordinate system; Dynamic adjustment of the high-resolution camera: according to the initial pixel coordinates, the rotatable high-resolution camera is controlled to adjust the pitch angle β and the yaw angle α around the optical axis so that the target enters its field of view; Multi-view image acquisition and modeling: Capture high-resolution images at different rotation angles and perform distortion correction to obtain the pixel coordinates of the target in the pixel coordinate system of the high-resolution camera; Multi-view geometric constraint construction: Based on the normalized coordinates of the wide-angle camera and the dynamic geometric relationship of the high-resolution camera, a nonlinear equation group including the pitch angle β and the yaw angle α is established; Absolute distance solution: Solve the nonlinear equations by numerical analysis method and output the absolute distance Z from the target to the wide-angle camera. p .

9. The target distance measurement method based on multi-angle camera collaborative perception according to claim 8, characterized in that: The construction of the multi-view geometric constraints includes: According to the pitch angle β and the yaw angle α, the dynamic projection relationship between the target point and the optical axis of the high-resolution camera is established; Combine the baseline distance b and the rotation angle parameter to generate the spatial distance and angle constraint equation; The numerical analysis method is Newton's iteration method, which achieves convergent solution of nonlinear equations through parameter matching.

10. The target ranging method based on multi-angle camera collaborative perception according to claim 8 or 9, characterized in that: When the target is within the field of view of the wide-angle camera, the steps of dynamically adjusting the high-resolution camera to the absolute distance solution are repeated, and the distance measurement results are updated through the multi-view geometric constraints of the high-resolution camera at different rotation angles.

Images