Ship height binocular measurement method for bridge ship collision risk early warning

Abstract

The invention discloses a ship height binocular measurement method for bridge ship collision risk early warning, and the method specifically comprises the steps: building a binocular vision measurement system facing a ship navigation region, installing two camera devices at the two sides of a navigation channel, and forming a binocular imaging structure with parallel optical axes; system calibration is completed through displacement adjustment and distance measurement, and a mapping relation between a world coordinate system and an image pixel coordinate system is established; the system obtains ship image information and extracts key point position information of a ship target in an image; image coordinates are converted into space coordinates, and the three-dimensional position and structural features of the ship are obtained; calculating main size parameters such as the height of the ship; and comparing the ship size data obtained based on measurement with a preset bridge clearance limit in a navigation area, and carrying out early warning on the ultrahigh ship. The method is widely applied to the field of bridge traffic safety, the active recognition and response capability of the bridge to the ship collision risk is improved, and stable and reliable technical support is provided for bridge navigation safety.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of bridge active anti-collision, and particularly relates to a ship height binocular measurement method for bridge ship collision risk early warning. BACKGROUND

[0002] With the rapid development of water transportation, the ship passing density in the navigation area increases significantly, and the tonnage of the ship also shows a significant increasing trend, which makes the risk of ship-bridge collision rise continuously, and the bridge on the navigation river is hit by the ship from time to time.

[0003] The current ship size measurement methods mainly include laser radar, monocular vision, binocular vision and the like. However, the laser radar faces the problem of failure in rainy and snowy weather, which can cause the over-limit ship not to be detected, and further cause the collision risk. The measurement accuracy of monocular vision is significantly affected by the distance, and the shooting of the ship image is carried out at a distance far from the bridge, which causes the measurement accuracy to be greatly affected. The binocular vision technology provides a new idea for solving the above problems. However, the research on the binocular vision method in the field of bridge anti-ship collision is still blank, and it is urgent to study the ship size measurement method from the perspective of binocular vision to improve the active anti-ship collision ability of the bridge. SUMMARY

[0004] Technical problem: The purpose of the application is to provide a ship height binocular measurement method for bridge ship collision risk early warning, which does not depend on high-density labeled data and can be widely applied in the field of bridge navigation safety, and provides reliable support for the active identification and response ability of the bridge to the ship collision risk.

[0005] Technical scheme: In order to achieve the above purpose, the ship height binocular measurement method for bridge ship collision risk early warning solution of the application includes the following steps: Step 1, a binocular vision measurement system is built, industrial cameras are installed on both sides of the ship navigation area and are calibrated, the internal and external parameters of the industrial cameras are obtained, and the conversion relationship between the world coordinate system and the industrial camera pixel coordinate system is established; Step 2, the ship image is collected by the industrial camera, the YOLOv8 model of fusion Transformer and deformable convolution DC is used for ship image instance segmentation, the edge consistency and filter adaptability are improved by combining direction perception Gaussian filter, structure residual adjustment and semantic attention mechanism, and the window size prediction based on local texture complexity and multi-scale pyramid structure are introduced, the dynamic adjustment of the filter window and the noise suppression are realized, on this basis, the edge detection is carried out, the target edge information is extracted, and finally the ship contour image with continuous structure and clear edge is output; Step 3, based on the calibration parameters of the binocular vision system and the ship feature point information, the pixel coordinates of the ship feature points are converted into the world coordinates, and the ship height is calculated according to the world coordinates of the ship feature points and the height of the ship feature points in the world coordinate system; u mp Convert to three-dimensional space coordinates in the world coordinate system X , Y , Z ), by calculating the key points of the ship in the three-dimensional space coordinates in the world coordinate system , determine its height, length and width, and finally based on the preset maximum clearance height of the channel, the ships with height exceeding the standard are warned.

[0006] Among them, In step 1, a binocular vision measurement system is built, industrial cameras are installed in the ship navigation area, and the camera internal and external parameters are calibrated, including left camera focal length , right camera focal length , principal point coordinates , rotation matrix and translation vector , wherein the left camera focal length , the right camera focal length are parameters describing the optical characteristics of the camera lens, the principal point coordinates is the coordinates of the intersection of the optical axis and the image plane in the image coordinate system, the nine elements R , a , b , c , d , e , , , , in the rotation matrix X , Y , Z represent the direction vectors of the three coordinate axes T , , in the right camera coordinate system in the coordinate system, the translation vector , X , Y represents the translation amount of the right camera relative to the left camera in the world coordinate system Z , s , , the scale factor is the pixel coordinate normalization, and the calibration process is as follows: , wherein represents the pixel unit length of the focal length along the horizontal axis of the image, i.e. axis direction,

[0007] In step 2, the ship image is collected by an industrial camera, and a Transformer module is embedded in the detection head based on the YOLOv8 architecture to enhance the global context of the feature map, thereby improving the detection accuracy of dense targets and occluded targets. The multi-head attention mechanism in the Transformer module is used to extract spatial dependent features of different semantic regions in the image in parallel. The deformable convolution module is introduced to enhance the modeling capability of the model for target deformation and complex boundaries.

[0008] In step 2, an adaptive composite filtering method is used for edge detection to remove noise. A local structure direction field is introduced in the filtering process. The covariance matrix of the Gaussian filter kernel is adjusted based on the direction field. The principal axis direction of the Gaussian kernel is consistent with the structure direction field, thereby improving the directional consistency of the filter in the structure area. , In the formula, is the covariance matrix of a two-dimensional Gaussian kernel at the image pixel position is the principal direction corresponding to the structure direction field is a two-dimensional rotation matrix that rotates the Gaussian kernel to the direction and represent the standard deviations in the parallel and perpendicular directions, respectively, controlling the extension of the filter kernel in the two directions. is the transpose matrix of Next, a structure residual adjustment branch is embedded. The edge response difference between the pixel value of the original image at position and the pixel value of the predicted image at position is minimized to enhance the fidelity of the image structure. The optimization objective function is: , In the formula,​​​ denotes traversing all coordinate positions in the image , the error of each pixel is accumulated, N denotes the total number of edge points participating in the calculation; denotes the Euclidean norm; E denotes the edge response fidelity error; In the filtering process, the semantic distribution map of the image is extracted based on a lightweight semantic segmentation network , a semantic attention mask is constructed , which is used to dynamically adjust the filtering strength , the adjustment process is defined as: , In the formula, is the filtering strength at position , is the basic filtering strength, denotes the function used to map the semantic score to the intensity weight; A structure confidence factor is introduced for each pixel , which is calculated by the edge intensity and the semantic boundary, and is used to weight and integrate the filtering output, and the output result of the final filtered image is expressed as: , In the formula, denotes the pixel value of the original image at position , denotes the pixel value of the image after filtering at position .

[0009] In step 2, the ship image is collected by an industrial camera, and a window size prediction module based on local texture complexity is introduced in the image processing process to realize adaptive adjustment of the filtering window, which specifically includes the following steps: By calculating the local gray variance or edge density of the neighborhood of each pixel in the image, the local texture complexity is evaluated, and the optimal size of the filtering window is dynamically predicted , the window size prediction function is defined as: , In the formula, and are weight coefficients, and are the maximum values of the gray variance and edge density of the entire image, respectively, for normalization processing; A multi-scale pyramid structure is designed to construct a pyramid image sequence including multiple resolution levels for an input image A filter window of a corresponding scale is applied on each scale level respectively Feature extraction and noise suppression are performed.

[0010] The binocular vision system calibration parameters and the ship feature point information in step 3 are measured by using a binocular stereo vision structure with collinear optical axes O-XYZ The left camera is installed at the origin of the world coordinate system, and no rotation transformation is applied R The right camera is converted to the world coordinate system by a rotation matrix T and a translation vector to establish a binocular stereo vision structure with collinear optical axes The left and right imaging plane optical centers are respectively, and the focal lengths are O-XYZ The key point has a corresponding relationship between the two image planes, which is as follows: , where s is the pixel coordinate normalization, represents the normalized coordinates of the spatial point on the left image, represents the normalized coordinates in the corresponding right image, and represents the depth value of the spatial point in the left camera coordinate system. Since the two image planes are mirror symmetric, the following equation is used: , Therefore, the three-dimensional coordinates of are simplified as: , where B is the distance between the two cameras, also known as the baseline.

[0011] The binocular vision system calibration parameters and the ship feature point information in step 3 further include the depth of the measured object in the Z-axis direction of the world coordinate system In the calibrated binocular vision measurement model, the height of the object in the 3D scene is calculated by using the focal length, baseline B and the coordinates of the highest point in the two images; the imaging heights of the points in the left and right image planes are and respectively, and the following triangulation is used in the binocular vision measurement system with collinear optical axes: , wherein, are the distances between the ship and the left and right cameras respectively; At the same time , B The data is obtained through calibration, and the formula for calculating the height of the ship is simplified as: .

[0012] Beneficial effects: Compared with the prior art, the present application has the following beneficial effects: High measurement accuracy and strong adaptability: The present application adopts a binocular stereo vision measurement system with collinear optical axes, combined with high-precision calibration parameters and a spatial geometric model, which can accurately calculate the three-dimensional coordinate position of the ship in the world coordinate system, significantly improving the accuracy of height measurement and avoiding errors caused by scale uncertainty in monocular vision measurement.

[0013] Strong edge detail retention capability: The YOLOv8 model with fusion Transformer and deformable convolution is introduced to realize high-precision instance segmentation, and combined with various edge enhancement strategies such as direction perception Gaussian filtering and structure residual adjustment, the edge consistency and image fidelity of the structure region are effectively improved, ensuring clear ship boundary and continuous contour.

[0014] Noise suppression and adaptive filtering adjustment: By introducing local texture complexity analysis and multi-scale pyramid structure, dynamic adaptive adjustment of the filter window size is realized, and further combined with semantic attention mask and structure confidence factor, different filtering strategies are executed for different regions, so as to balance the noise reduction ability and edge retention performance of the image.

[0015] Strong practicality of bridge anti-collision warning: The present application does not need to rely on high-density labeled data, and the measurement process is highly automated, suitable for real-time height measurement and warning judgment of different types of ships in complex navigation environment, which can provide reliable perception basis and warning basis for the active anti-collision system of the bridge, reduce the risk of bridge structure collision, and improve the navigation safety.

[0016] In summary, the present application not only has innovation in image processing algorithm, but also has good practicality and popularization value in engineering deployment, which can provide effective support for bridge structure safety monitoring and ship traffic management. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a flowchart of the present application; Figure 2 is a schematic diagram of the binocular vision system; Figure 3 is an instance segmentation diagram; Figure 4 is a multi-scale pixel pyramid diagram; Figure 5 This is a schematic diagram of binocular imaging with collinear optical axes. Detailed Implementation

[0018] The technical solution and beneficial effects of the present invention will be described in detail below with reference to the accompanying drawings.

[0019] like Figure 1 As shown, the present invention provides a binocular method for ship height measurement for bridge collision risk early warning, comprising the following steps: Step 1: Set up a binocular vision measurement system, install industrial cameras on both banks of the ship navigation area and calibrate them. Figure 2 (This involves) acquiring the intrinsic and extrinsic parameters of the industrial camera and establishing the transformation relationship between the world coordinate system and the pixel coordinate system of the industrial camera; Step 2: Acquire ship images using an industrial camera, and then segment the ship images using a YOLOv8 model that fuses Transformer and Deformable Convolution (DC). Figure 3 This combines orientation-aware Gaussian filtering, structural residual adjustment, and semantic attention mechanisms to improve edge consistency and filtering adaptability, and introduces window size prediction based on local texture complexity and a multi-scale pyramid structure. Figure 4 This allows for dynamic adjustment of the filtering window and noise suppression; finally, it outputs a ship outline image with continuous structure and clear edges. Step 3: Based on the calibration parameters of the binocular vision system and the ship's feature point information, the pixel coordinates are... Convert to three-dimensional space coordinates in the world coordinate system ( X , Y , Z By calculating the ship's critical points Three-dimensional spatial coordinates in the world coordinate system The height, length, and width of the vessel are determined; finally, warnings are issued to vessels exceeding the height limit based on the maximum clearance height preset for the waterway.

[0020] Furthermore, the method for constructing the binocular vision measurement system in step one is as follows: Figure 2 As shown, industrial-grade CMOS cameras (model Basler acA2440-35um) were installed on both banks of the ship navigation area to form a binocular stereo vision system with a fixed baseline length of B=1.460km. The lens focal length was set to... =24mm, image resolution is 2448×2048 pixels. Perform camera intrinsic and extrinsic parameter calibration to obtain the following parameters; principal point coordinates: Rotation matrix With translation vector Establish a world coordinate system ( X , Y ,Z ) and pixel coordinate system The conversion relationship between them. Among them, the focal lengths of the left and right cameras are parameters describing the optical characteristics of the camera lenses. These are the coordinates of the intersection point of the optical axis and the image plane in the image coordinate system, and the rotation matrix. R The translation vector represents the rotation relationship between the right camera coordinate system and the left camera coordinate system. T Describe the translational displacement of the origin of the right camera coordinate system relative to the origin of the left camera coordinate system; Furthermore, in step one, the camera calibration process is as follows: , In the formula, s Used for pixel coordinate normalization The focal length is represented along the horizontal axis of the image. The length of a pixel unit along the axial direction. It represents the pixel unit length of the focal length along the vertical axis of the image, i.e., the v-axis.

[0021] like Figure 3 As shown, in step two, based on the YOLOv8 architecture, a Transformer module is embedded in the detection head to perform global context enhancement processing on the feature map, thereby improving the model's detection accuracy for dense and occluded targets. The multi-head attention mechanism in the Transformer module is used to extract spatial dependency features of different semantic regions in the image in parallel. The multi-head attention has a dimension of 8 and a nesting layer of 2 to enhance the spatial dependency modeling ability of densely occluded regions. A deformable convolution module is then introduced to enhance the model's ability to model target deformation and complex boundaries. This module introduces a spatial offset at each output position's convolution sampling point to achieve... Dynamic adjustment of the convolutional region allows it to adapt to feature extraction requirements under different structural forms. The backbone convolutional layer kernel is replaced, and the sampling offset dimension is... This improves the responsiveness to non-rigid boundaries. Each sampling point learns its position information independently, and all channels within a batch share the same set of offsets to ensure computational consistency and reduce redundancy.

[0022] Furthermore, in step two, the edge extraction process in step 2 employs an adaptive composite filtering method to remove noise. A local structural orientation field is introduced during the filtering process. The covariance matrix of the Gaussian filter kernel is adjusted based on this directional field. This ensures that the principal axis of the Gaussian kernel aligns with the structural orientation field, thereby improving the directional consistency of the filter in the structural region. The covariance matrix is ​​defined as follows: , In the formula, A two-dimensional Gaussian kernel at the pixel position in the image The covariance matrix at that location, For structural direction field The corresponding main direction, To rotate the Gaussian kernel to the direction Two-dimensional rotation matrix, and These represent the standard deviations in the parallel and perpendicular directions, respectively, controlling the extent to which the filter kernel expands in both directions. Take 2.0, Take 0.5.

[0023] Subsequently, a structural residual adjustment branch is embedded, and the original image is positioned... pixel value at and predicted image at location pixel value at Edge response differences To minimize the image and enhance the fidelity of its structure, the objective function is: , In the formula, This indicates traversing all coordinate positions in the image. The error of each pixel is summed up. N This represents the total number of edge points involved in the calculation; This represents the Euclidean norm.

[0024] Then, the semantic distribution map of the image is extracted based on the lightweight semantic segmentation network (SegFormer-B0). Constructing semantic attention masks Used for dynamically adjusting filter strength The adjustment process is defined as follows: , In the formula, For in position Filtering intensity at that point The base filter strength is set to 1.2. express A function used to map semantic scores to intensity weights; Finally, a structure confidence factor is introduced for each pixel. This factor is calculated by combining edge strength and semantic boundary, and is used to weight and integrate the filtered output, resulting in the final filtered image output. Expressed as: , In the formula, Indicates the original image at position Pixel value at that location, This indicates the position of the image after filtering. The pixel value at that location.

[0025] Furthermore, step 2 of the image processing also introduces a window size prediction module based on local texture complexity. This module assesses texture complexity by calculating the local gray-level variance and edge density of the pixel neighborhood, thereby dynamically predicting the filter window size. The window size prediction function is defined as follows: , In the formula, This is the optimal size for the filtering window. Let V be the local gray-level variance of each pixel's neighborhood in the image. For edge density, and These are the weighting coefficients, with values ​​of [values ​​to be filled in]. , , and These are the maximum values ​​of the overall grayscale variance and the edge density, respectively, used for normalization.

[0026] like Figure 4 As shown, the multi-scale pyramid structure constructs a pyramid image sequence with multiple resolution levels from the input image. Construct a 5-layer pyramid here, with each layer having a resolution equal to that of the original image. Apply the corresponding scale filtering window at each scale level. Feature extraction and noise suppression are performed.

[0027] Furthermore, the binocular measurement system described in step 3 is a three-dimensional structure with collinear optical axes, such as... Figure 5 The industrial camera on the left is fixed in the world coordinate system. O-XYZ Origin, without applying any rotation transformation, rotation matrix between the right industrial camera and it. R With translation moment T Represented as: , In the formula, there are nine elements. a b, c, d, e , , , This represents the three coordinate axes in the right camera coordinate system. X , Y , Z The direction vector has three components in this coordinate system, which are the components of the rotation matrix. , , This indicates the position of the right camera relative to the left camera in the world coordinate system. X、 Y 、 Z translation in x and y directions, collectively called translation vector T ; the optical centers of left and right imaging planes are , and the focal lengths are . A spatial point O-XYZ has a corresponding relationship between the two image planes, which is as follows: , where s is the pixel coordinate normalization, denotes the normalized coordinate of the spatial point on the left image, denotes the normalized coordinate in the corresponding right image, denotes the depth value of the spatial point in the left camera coordinate system. The three-dimensional coordinates of , where B is the distance between the two cameras, also known as the baseline.

[0028] Further, step 3 also includes the calculation of the position depth of the object in the Z-axis direction. Through the highest point of the ship in the image , the imaging heights of the left and right images are and , combined with the binocular geometric calibration parameters, the object height is calculated using the following triangular relationship: , where are the distances between the ship and the left and right cameras, respectively.

[0029] At the same time , B the data of can be obtained through calibration, and the formula for calculating the height of the ship can be simplified as: .

Claims

1. A ship height binocular measurement method for bridge ship collision risk early warning, characterized in that Comprising the following steps: Step 1, build a binocular vision measurement system, install industrial cameras on both sides of the ship navigation area and calibrate to obtain the internal and external parameters of the industrial cameras, and establish the conversion relationship between the world coordinate system and the industrial camera pixel coordinate system; Step 2, collect ship images through industrial cameras, use the YOLOv8 model combining fusion Transformer and deformable convolution DC for ship image instance segmentation, combine direction perception Gaussian filtering, structure residual adjustment and semantic attention mechanism to improve edge consistency and filtering adaptability, and introduce window size prediction based on local texture complexity and multi-scale pyramid structure to realize dynamic adjustment of the filtering window and noise suppression, and then carry out edge detection to extract target edge information, and finally output the ship contour image with continuous structure and clear edge; Step 3: Based on the calibration parameters of the binocular vision system and the ship's feature point information, the pixel coordinates are... Convert to three-dimensional space coordinates in the world coordinate system ( X , Y , Z By calculating the ship's critical points Three-dimensional spatial coordinates in the world coordinate system The height, length, and width of the vessel are determined, and finally, based on the maximum clearance height preset for the waterway, warnings are issued to vessels exceeding the height limit.

2. The ship height binocular measurement method for bridge ship collision risk early warning according to claim 1, characterized in that, In the step 1, a binocular vision measurement system is built, industrial cameras are installed in the ship navigation area, and the internal and external parameters of the cameras are calibrated, including left camera focal length , right camera focal length , principal point coordinates , rotation matrix and translation vector , wherein the left camera focal length , the right camera focal length are parameters describing the optical characteristics of the camera lens, the principal point coordinates are the coordinates of the intersection of the optical axis and the image plane in the image coordinate system, the nine elements R , a , b , c , d , e , , , , of the rotation matrix R represent the three components of the direction vectors of the three coordinate axes X , Y , Z in the right camera coordinate system in the coordinate system, and the translation vector T , , , represents the translation amount of the right camera relative to the left camera in the world coordinate system X , Y , Z , and the scale factor s is the pixel coordinate normalization, and the calibration process is as follows: ， In the formula, represents the pixel unit length of the focal length along the image horizontal axis, i.e. the axis direction, represents the pixel unit length of the focal length along the image vertical axis, i.e. the v-axis direction.

3. The binocular measurement method of ship height for bridge ship collision risk early warning according to claim 2, characterized in that, In step 2, the ship image is collected through the industrial camera, which is based on the YOLOv8 architecture, and the Transformer module is embedded in the detection head to enhance the global context of the feature map, thereby improving the detection accuracy of dense targets and occluded targets; The multi-head attention mechanism in the Transformer module is used to extract spatial dependent features of different semantic regions in the image in parallel; A deformable convolution module is introduced to enhance the model's ability to model target deformation and complex boundaries. The module introduces spatial offsets for each output position's convolution sampling point to achieve Dynamic adjustment of the convolution region enables it to adapt to feature extraction requirements under different structural forms. Each sampling point independently learns position information, and all channels in a batch share the same set of offsets to ensure computational consistency and reduce redundancy. wherein denotes the size of the convolution kernel, which describes the size of the sampling region in the convolution operation.

4. The ship height binocular measurement method for bridge ship collision risk early warning according to claim 3, characterized in that, In step 2, the edge detection uses an adaptive composite filtering method to remove noise, which includes the following processing process: Introducing a local structure direction field in the filtering process and adjusting the covariance matrix of the Gaussian filter kernel according to the direction field such that the principal axis of the Gaussian kernel is aligned with the structure direction field, thereby improving the directional consistency of the filtering in the structure region, the covariance matrix being defined as follows: ， wherein is a covariance matrix of a two-dimensional Gaussian kernel at image pixel position , is the principal direction corresponding to the structure direction field , is a two-dimensional rotation matrix that rotates the Gaussian kernel to the direction , and denote the standard deviations in the parallel and perpendicular directions, respectively, controlling the extent of the filter kernel in the two directions; is the transpose of . Then the structure residual adjustment branch is embedded, the edge response difference of the pixel value of the original image at position and the pixel value of the predicted image at position is minimized to enhance the fidelity of the image structure, and the optimization objective function is: ​​​ ， wherein represents traversing all coordinate positions in the image accumulating the error of each pixel, N represents the total number of edge points participating in the calculation; represents the Euclidean norm; E represents the edge response fidelity error; In the filtering process, a semantic distribution map of the image is extracted based on a lightweight semantic segmentation network , a semantic attention mask is constructed , the filtering strength is dynamically adjusted , and the adjustment process is defined as: ， wherein is the filter strength at position , is the base filter strength, denotes a function to map semantic scores to intensity weights; Introducing a structure confidence factor for each pixel The factor is calculated by the edge strength and semantic boundary, which is used to weight the integration of the filtering output, and the output result of the final filtering image Expressed as: ， wherein represents a pixel value of the original image at position represents a pixel value of the original image at position represents a pixel value of the filtered image at position represents a pixel value of the filtered image at position 5. The ship height binocular measurement method for bridge ship collision risk early warning according to claim 4, characterized in that, In step 2, the ship image is collected through the industrial camera, and a window size prediction module based on local texture complexity is introduced in the image processing process to realize adaptive adjustment of the filtering window, which includes the following steps: by calculating the local gray level variance of the neighborhood of each pixel in the image or the edge density , the local texture complexity is evaluated, based on which the optimal size of the filtering window is dynamically predicted , the window size prediction function is defined as: ， In the formula, With is a weight coefficient, and respectively, the maximum value of the full image gray scale variance and the edge density, for normalization processing; A multi-scale pyramid structure is designed to construct a pyramid image sequence including multiple resolution levels from an input image A filter window of a corresponding scale is applied at each scale level, respectively Feature extraction and noise suppression are performed.

6. The binocular measurement method of ship height for bridge ship collision risk early warning according to claim 5, characterized in that, The calibration parameters based on the binocular vision system described in step 3 and the ship feature point information are a binocular stereo vision structure with collinear optical axes O-XYZ The left camera is installed at the origin of the world coordinate system, and no rotation transformation is applied, and the right camera is converted to the world coordinate system through a rotation matrix R and a translation vector T , so as to establish a binocular stereo vision structure with collinear optical axes; The optical centers of the left and right imaging planes are respectively , and the focal lengths are respectively ; Located O-XYZ Key points Between the two image planes have a corresponding relationship, as follows: ， wherein s is the pixel coordinate normalization, denotes the normalized coordinate of the spatial point on the left image, denotes the normalized coordinate in the corresponding right image, denotes the depth value of the spatial point in the left camera coordinate system; Since the two image planes are mirror symmetrical, The following equation is expressed: ， Thus, The three-dimensional coordinates of the point of intersection of the line and the plane are simplified to: ， In the formula, B is the distance between the two cameras, also called the baseline.

7. The binocular measurement method of ship height for bridge ship collision risk early warning according to claim 6, characterized in that, The calibration parameters of the binocular vision system and the ship feature point information further include the depth of the position of the measured object in the Z-axis direction of the world coordinate system In the calibrated binocular vision measurement model, the height of the object in the 3D scene is calculated by using the focal length, the baseline B and the coordinates of the highest point in the two images The imaging heights of the points in the left and right image planes are and In the binocular vision measurement system with collinear optical axes, the following triangulation is used: ， In the formula, respectively are the distances between the ship and the left and right cameras. At the same time , B The data is obtained by calibration, and the formula for calculating the height of the ship is simplified as: 。

Images