Image total station visual corner measurement method and system for elevation angle adaptive calibration

Abstract

The invention provides an image total station visual corner measurement method and system for elevation angle adaptive calibration, and relates to the technical field of surveying and mapping engineering.The method comprises the steps that an initial elevation angle in a point position library of an image total station is obtained, measurement is conducted through angle matching and a calibration model, and a visual corner collection image is obtained; preprocessing the visual corner acquisition image to obtain enhanced image data; performing multi-type target centroid extraction on the enhanced image data to obtain centroid data; based on the centroid data, performing pixel angle coordinate conversion by using a calibration model to obtain a coordinate conversion result; performing multi-dimensional error correction based on the coordinate conversion result to obtain a correction result; and performing corner measurement and precision verification on the correction result to obtain a visual corner measurement result. According to the invention, the problems of complex environment adaptation defects of the traditional ATR technology and insufficient calibration precision and low automation degree of the existing image scheme are solved.

Description

Technical Field

[0001] This manual relates to the field of surveying and mapping engineering technology, and in particular to a method and system for measuring the visual edges and corners of an image total station with elevation angle adaptation calibration. Background Technology

[0002] Existing total station automatic measurement technology suffers from a dual bottleneck: on the one hand, traditional ATR technology is difficult to adapt to complex environments; on the other hand, the calibration and extraction accuracy of image-assisted schemes is insufficient. The inherent limitations of traditional ATR technology include the following: 1. Poor stability in strong light environments: ATR relies on 635nm / 650nm laser reflection signals. When the light intensity is strong, stray light from the environment causes the signal-to-noise ratio of the reflected light to drop from 30:1 to below 5:1, and the target recognition success rate drops sharply from 98% to 65%, even less than 30% under strong light, making continuous measurement impossible. 2. Weak adaptability to occluded scenes: The laser beam diameter is only 2-5mm (at 100m). If there is more than 10% occlusion in the measurement path, the intensity of the reflected signal decreases by 40%+, triggering measurement interruption; discontinuous occlusion (such as tree branches and leaves) will lead to a "aiming-losing-reaiming" cycle, extending the time for a single measurement to 5-10 minutes, and greatly increasing the horizontal angle error. 3. Missing measurement of non-prism targets: ATR requires the high reflectivity of prisms (reflectivity ≥90%), which is completely ineffective for non-prism targets (reflectivity 5%-30%, such as ancient building components and tunnel cracks); the method of manually pasting reflective sheets is inefficient (reduced by 50%+) and has poor safety, and some targets are prohibited from being pasted.

[0003] The technical shortcomings of existing image-assisted solutions include: 1. Semi-automation relying on manual labor: Existing solutions only use images for "visual-assisted aiming," requiring manual adjustment of the total station angle to align with the target. Essentially, it remains semi-automated, with limited improvement in efficiency per measurement compared to traditional manual methods. 2. Insufficient coordinate transformation accuracy: No polynomial transformation model adapted to elevation angles has been established, relying instead on a single linear or linear model; furthermore, the fitting residuals are difficult to stabilize. 3. Target extraction and efficiency deficiencies: The accuracy and applicability for handling different targets in complex situations are insufficient; non-prism target extraction relies solely on Canny edge detection, lacking specific optimization for regular circular targets (commonly used in engineering). 4. Incomplete error control: The system does not systematically integrate corrections for "systematic error - environmental error - random error." Angle errors caused by structural thermal expansion and contraction when the temperature changes by 10℃ are not compensated, resulting in deviations of over ±1.0″ for multiple measurements of the same target. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this invention provides a method and system for measuring the visual edges and corners of a total station using an image with elevation angle adaptation calibration. This solution addresses the complex environment adaptation defects of traditional ATR technology, the insufficient calibration accuracy of existing image schemes, and the low degree of automation.

[0005] To achieve the aforementioned objectives, the technical solution adopted by this invention is: a method for measuring the visual edge angle of an image total station with elevation angle adaptation calibration, comprising: S1: Obtain the initial elevation angle from the point database of the image total station, and measure it using the calibration model through angle matching to obtain the visual edge and corner acquisition image; S2: Preprocess the visually captured images at the edges and corners to obtain enhanced image data; S3: Extract centroids of multiple types of targets from the enhanced image data to obtain centroid data; S4: Based on the centroid data, use the calibration model to perform pixel angle coordinate transformation and obtain the coordinate transformation result; S5: Based on the coordinate transformation results, perform multi-dimensional error correction to obtain the correction results; S6: Perform edge and corner measurement and accuracy verification on the correction results to obtain the visual edge and corner measurement results, and complete the visual edge and corner measurement of the image total station.

[0006] Further, S1 includes: The initial elevation angle is obtained from the point database of the total station. When the elevation angle is less than or equal to 30°, the low elevation angle model is used, and when the elevation angle is greater than 30°, the high elevation angle model is used. When the distance is less than 300m, a 12mm focal length is used, and when the distance is between 300m and 600m, a 200mm focal length is used. The exposure time is no more than 1 / 500s in strong light and no more than 1 / 10s in weak light. Measurements are taken to obtain visual edge and corner acquisition images. The low elevation angle model and the high elevation angle model are calibration models.

[0007] Furthermore, the calibration model also includes a pixel-angle nonlinear relationship, and the expression for the low elevation angle model is: ; The expression for the high elevation angle model is: ; The expression for the pixel-angle nonlinear relationship is: ; in, This indicates the low elevation angle proportional compensation coefficient. This represents the high altitude angle proportional compensation coefficient. , , , , , and Indicates the horizontal angle conversion factor. Indicates horizontal pixels, Represents vertical pixels. Indicates the real-time elevation angle. , , , , and This represents the elevation angle conversion factor.

[0008] Further, S2 includes: The visual corner acquisition images are processed by region acquisition to obtain multi-region visual corner acquisition images; A filter kernel is used to perform joint denoising on visual corner acquisition images of multiple regions through Gaussian filtering and median filtering to obtain the denoising result; The denoising results are then subjected to grayscale conversion and contrast enhancement to obtain enhanced image data.

[0009] Further, S3 includes: For prism targets with high reflectivity in enhanced image data, adaptive threshold segmentation, morphological noise suppression, and minimum circumcircle fitting localization are processed sequentially to obtain the centroid pixel coordinates of the prism target with the center of the fitted circle as the centroid pixel coordinates. For regular circular targets in enhanced image data, the global centroid pixel coordinates of the regular circular targets are obtained by sequentially processing the black substrate region segmentation and optimization, the circularity screening of the substrate outline, and the positioning and optimization of the white center. For irregular non-prism targets in enhanced image data, an edge detection algorithm is used to extract the edge contours of the targets by setting a dual threshold range. The contour length is then filtered to obtain the geometric moments of the irregular non-prism targets. Among them, the centroid pixel coordinates of the prism targets, the global centroid pixel coordinates of the regular circular targets, and the geometric moments of the irregular non-prism targets are considered centroid data, with the center of the fitted circle as the centroid pixel coordinates.

[0010] Further, S5 includes: Based on the coordinate transformation results, systematic error correction, environmental error correction, and random error correction are performed respectively to obtain systematic error correction results, environmental error correction results, and random error correction results; among them, systematic error correction results, environmental error correction results, and random error correction results are correction results.

[0011] Furthermore, the expression for the system error correction result is as follows: ; ; The expression for the environmental error correction result is: ; ; ; ; in, This indicates the horizontal angle after only systematic error correction. This represents the horizontal angle observation value before correction. Indicates the systematic error of the horizontal angle. This indicates the elevation angle after only systematic error correction. This represents the elevation angle observation value before correction. Indicates the systematic error of the elevation angle. This represents the environmental error amount of the horizontal angle. Indicates the ambient temperature. Indicates atmospheric pressure. This represents the final horizontal angle after system and environmental error correction. This represents the environmental error amount of the horizontal angle. This represents the final elevation angle after system and environmental error correction. This represents the environmental error amount of the elevation angle.

[0012] Furthermore, the visual angle measurement results include horizontal distance and vertical distance, and the expressions for horizontal distance and vertical distance are as follows: ; ; in, Indicates horizontal distance. Indicates the distance to the target. Indicates the elevation angle. Indicates vertical distance.

[0013] This invention provides an image total station visual edge and angle measurement system with elevation angle adaptation calibration, comprising: The main body of the engineering survey includes the total station host: horizontal angle encoder, elevation angle encoder, distance measuring module and optical path system, which are used to acquire the initial elevation angle in the point library of the image total station; The embedded imaging unit, including a telephoto CMOS module and an atmospheric scattering filter, is used to obtain visual edge and corner acquisition images by measuring using a calibration model through angle matching. The automatic control unit, including a processor and a dedicated image processing chip, is used to control focus switching, imaging parameters, measurement process triggering, and calibration model matching. The data processing unit includes a calibration model for elevation angle adaptation, a multi-target extraction algorithm, target pixel range prediction logic, and an engineering target point library. It is used to preprocess visually acquired corner images to obtain enhanced image data; extract centroids of multiple target types from the enhanced image data to obtain centroid data; perform pixel angle coordinate transformation using the calibration model based on the centroid data to obtain coordinate transformation results; and perform corner measurement and accuracy verification on the correction results to obtain visual corner measurement results, thus completing the visual corner measurement of the image total station. The environmental sensing module, including a temperature sensor, a barometric pressure sensor, and an elevation angle acquisition module, is used to collect environmental parameters, providing a basis for error correction and model calibration.

[0014] The beneficial effects of this invention are as follows: This invention provides a total station visual angle measurement method with elevation angle adaptation calibration. It replaces the traditional ATR (Automatic Target Recognition) laser aiming with visual technology, and improves the pixel-angle conversion accuracy of the total station by combining a low / high elevation angle segmented calibration model. It also integrates the Hough transform-center fitting extraction of a regular circular target with a closed-loop logic of manual initial calibration-automatic retesting, thereby achieving fully automatic angle measurement with third-order accuracy in short-to-medium distance engineering scenarios.

[0015] (1) Significantly improved adaptability to complex environments: The pure vision measurement scheme replaces the traditional ATR laser aiming, effectively solving the measurement failure problem in strong light and occlusion scenarios, while achieving accurate measurement of non-prism targets, greatly expanding the range of environmental adaptability; (2) The accuracy of full elevation angle measurement has been greatly optimized: By using the low / high elevation angle segmented calibration model, the influence of elevation angle on coordinate transformation is significantly reduced, the measurement accuracy is stable and meets the requirements of high-precision engineering specifications, and the data reliability is significantly enhanced; (3) The degree of automation and processing efficiency of measurement are significantly improved: After manual initial calibration, no intervention is required throughout the process, realizing a fully automatic edge and corner measurement process. At the same time, the useless data processing is reduced by predicting the target pixel range, and the overall measurement and data processing efficiency is greatly improved. (4) The target compatibility and scene adaptability are greatly enhanced: it can be adapted to various target types such as prism, regular circle, and irregular non-prism, and is widely applicable to engineering surveying, cultural relic mapping, deformation monitoring and other fields, and its practical value is significantly improved. Attached Figure Description

[0016] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein: Figure 1This is a schematic diagram illustrating an application scenario of an image total station visual edge and corner measurement system with elevation angle adaptation calibration, as shown in some embodiments of this specification. Figure 2 This is an exemplary flowchart illustrating an image total station visual edge angle measurement method with elevation angle adaptation calibration according to some embodiments of this specification. Detailed Implementation

[0017] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0018] Example 1 Figure 1 This is a schematic diagram of a total station visual edge and corner measurement system with elevation angle adaptation calibration, as shown in some embodiments of this specification.

[0019] In some embodiments, the image total station visual edge and corner measurement system with elevation angle adaptation calibration may include an engineering survey body, an embedded imaging unit, an automatic control unit, a data processing unit, and an environmental sensing module.

[0020] In some embodiments, an image total station visual edge angle measurement system with elevation angle adaptation calibration can solve the problems of accuracy attenuation and single verification in traditional calibration through three steps: "indoor control point calibration + low / high elevation angle optimization + simulation verification". Indoor control point basic calibration (β≤30°, low elevation angle model): Calibration environment setup: ≥8 static measurement calibration points (accuracy ±5mm) are set up in the indoor calibration field, covering elevation angles of 5°-30°, ensuring coverage of common elevation angle ranges in short-range engineering measurements; Data acquisition: The image total station is controlled to sequentially aim at each calibration point, synchronously acquiring: the pixel coordinates of the calibration point (u... i v i (u = horizontal pixel, v = vertical pixel, the origin of the pixel coordinate system is the top left corner of the image, u-axis to the right, v-axis to the bottom); Total station measured angle coordinates (α) i =Horizontal angle, β i =Elevation angle).

[0021] The main body of the engineering surveying system includes the total station host: horizontal angle encoder, elevation angle encoder, distance measuring module, and optical path system, used to acquire the initial elevation angle from the total station's point database. The main body of the engineering surveying system forms the foundation for engineering surveying from 0-600m, providing raw angle / distance data.

[0022] In some embodiments, a horizontal angle encoder (accuracy ≥ ±0.5″), an altitude angle encoder (accuracy ≥ ±0.5″), a ranging module (±(2mm+2ppm×D)); and an optical path system (parallelism error between the optical axis of the imaging module and the measuring axis ≤ 0.005°).

[0023] The embedded imaging unit, including a long-focus CMOS module and an atmospheric scattering filter, is used to obtain visual edge-captured images through angle matching and measurement using a calibration model. The embedded imaging unit is used for high-target image acquisition; 12mm is suitable for <300m, and 200mm is suitable for 300-600m. The filter is resistant to haze interference.

[0024] In some embodiments, a telephoto CMOS module (resolution ≥1920×2160, focal length 12mm-infinity); an atmospheric scattering filter (automatically switched).

[0025] The automatic control unit, including a quad-core processor and a dedicated image processing chip, is used to control focus switching, imaging parameters, measurement process triggering, and calibration model matching. The automatic control unit controls focus switching, imaging parameters (exposure 1 / 10s-1 / 1000s), measurement process triggering, and model matching.

[0026] The data processing unit includes a calibration model for elevation angle adaptation, a multi-target extraction algorithm, target pixel range prediction logic, and an engineering target point library. It is used to preprocess visual corner acquisition images to obtain enhanced image data; extract centroids of multiple types of targets from the enhanced image data to obtain centroid data; perform pixel angle coordinate transformation based on the centroid data using the calibration model to obtain coordinate transformation results; and perform corner measurement and accuracy verification on the correction results to obtain visual corner measurement results, thus completing the visual corner measurement of the image total station.

[0027] The environmental sensing module, including a temperature sensor, a barometric pressure sensor, and an elevation angle acquisition module, is used to collect environmental parameters, providing a basis for error correction and model calibration.

[0028] In some embodiments, a temperature sensor (±0.5°C) and a barometric pressure sensor (±1 hPa) are used.

[0029] In some embodiments, an image total station visual corner measurement system with elevation angle adaptation calibration can be used to perform an image total station visual corner measurement method with elevation angle adaptation calibration, including: S1: acquiring the initial elevation angle from the point database of the image total station, measuring it using a calibration model through angle matching, and obtaining a visual corner acquisition image; S2: preprocessing the visual corner acquisition image to obtain enhanced image data; S3: extracting centroids of multiple types of targets from the enhanced image data to obtain centroid data; S4: performing pixel angle coordinate transformation using the calibration model based on the centroid data to obtain coordinate transformation results; S5: performing multi-dimensional error correction based on the coordinate transformation results to obtain correction results; S6: performing corner measurement and accuracy verification on the correction results to obtain visual corner measurement results, thus completing the visual corner measurement of the image total station.

[0030] In some embodiments of this specification, the processor utilizes an image total station visual corner measurement system with elevation angle adaptation calibration to perform an image total station visual corner measurement method with elevation angle adaptation calibration. In this way, (1) the adaptability to complex environments is significantly improved: the pure vision measurement scheme replaces the traditional ATR laser aiming, effectively solving the measurement failure problem in strong light and occlusion scenarios, while realizing the accurate measurement of non-prism targets, and greatly expanding the range of environmental adaptability; (2) the measurement accuracy of the full height angle is greatly optimized: through the low / high height angle segmented calibration model, the influence of the height angle on coordinate transformation is significantly reduced, the measurement accuracy is stable and meets the requirements of high-precision engineering specifications, and the data reliability is significantly enhanced; (3) the degree of measurement automation and processing efficiency are significantly improved: after manual initial calibration, no intervention is required throughout the process, realizing a fully automatic edge and corner measurement process, while reducing useless data processing through target pixel range prediction, and greatly improving the overall measurement and data processing efficiency; (4) the target compatibility and scene adaptability are greatly enhanced: it can be adapted to various target types such as prisms, regular circles, and irregular non-prisms, and is widely applicable to engineering surveying, cultural relic mapping, deformation monitoring and other fields, and its practical value is significantly enhanced.

[0031] Example 2 Figure 2 This is an exemplary flowchart illustrating an image total station visual edge angle measurement method with elevation angle adaptation calibration, according to some embodiments of this specification. Figure 2 As shown, the process includes the following steps. In some embodiments, the process may be executed by a processor.

[0032] S1: Obtain the initial elevation angle from the point database of the image total station, and measure it using the calibration model through angle matching to obtain the visual edge acquisition image.

[0033] Image total stations include prism targets, non-prism targets, and manual operation.

[0034] The prism target is a traditional measuring prism (reflectivity ≥ 90%).

[0035] Non-prism targets include regular circular targets (black substrate diameter 10-20cm, central white dot diameter 1-2cm, non-reflective) and irregular targets (such as stone corners of ancient buildings).

[0036] The manual operation involves setting up targets at the measurement points within a 0-600m range, manually aiming at the target center, recording initial parameters (α0 = horizontal angle, β0 = elevation angle, X0 / Y0 / Z0 = three-dimensional coordinates), associating the target type (prism / regular circle / irregular non-prism), and storing the data in the "Engineering Target Point Library" (supports storing ≥50 points). (This is used to remember the initial approximate coordinates for convenient subsequent repeated measurements).

[0037] Visual corner acquisition images are acquired images obtained by measuring using a calibration model after angle matching.

[0038] In some embodiments, the processor can acquire the initial elevation angle from the point library of the total station. When the elevation angle is less than or equal to 30°, a low elevation angle model is activated; when the elevation angle is greater than 30°, a high elevation angle model is activated. When the distance is less than 300m, a 12mm focal length is used; when the distance is between 300m and 600m, a 200mm focal length is used. The exposure time is no more than 1 / 500s in strong light and no more than 1 / 10s in weak light. Measurements are then taken to obtain visual edge-captured images. The low elevation angle model and the high elevation angle model are calibration models.

[0039] In some embodiments, the calibration model further includes a pixel-angle nonlinear relationship, and the expression for the low elevation angle model is: ; The expression for the high elevation angle model is: ; The expression for the pixel-angle nonlinear relationship is: ; in, This indicates the low elevation angle proportional compensation coefficient. This represents the high altitude angle proportional compensation coefficient. , , , , , and Indicates the horizontal angle conversion factor. Indicates horizontal pixels, Represents vertical pixels. Indicates the real-time elevation angle. , , , , and This represents the elevation angle conversion factor.

[0040] S2: Preprocess the visual corner acquisition images to obtain enhanced image data.

[0041] Enhanced image data is a visually captured image at the corners after filtering, grayscale conversion, and contrast enhancement.

[0042] In some embodiments, the processor can perform regional acquisition on the visual corner acquisition image to obtain multi-region visual corner acquisition images; use a filtering kernel to perform joint denoising on the multi-region visual corner acquisition images through Gaussian filtering and median filtering to obtain a denoising result; and perform grayscale conversion and contrast enhancement processing on the denoising result to obtain enhanced image data.

[0043] In some embodiments, joint denoising can be performed using a 3×3 / 5×5 filter kernel (5×5 for noise intensity > 50, otherwise 3×3) to perform Gaussian filtering (to remove Gaussian noise) and median filtering (to remove salt and pepper noise) on multi-region visual corner acquisition images to obtain the denoising result.

[0044] In some embodiments, grayscale can be achieved using a weighted average method, with grayscale value = 0.299R + 0.587G + 0.114B (R / G / B are color channel values); contrast enhancement can be achieved by limiting contrast adaptive histogram equalization, dividing the image into 8×8 blocks, with a gain of 2.0, to avoid distortion of the target edge caused by over-enhancement.

[0045] S3: Extract centroids of multiple target types from the enhanced image data to obtain centroid data.

[0046] In some embodiments, for a prism target with high reflectivity in the enhanced image data, adaptive threshold segmentation, morphological noise suppression, and minimum circumcircle fitting localization are performed sequentially to obtain the centroid pixel coordinates of the prism target with the center of the fitted circle as the centroid pixel coordinates. For a regular circular target in the enhanced image data, black substrate region segmentation and optimization, circularity filtering of the substrate contour, and localization and optimization of the white center are performed sequentially to obtain the global centroid pixel coordinates of the regular circular target. For an irregular non-prism target in the enhanced image data, an edge detection algorithm is used to extract the edge contour of the target by setting a dual threshold range, and the contour length is filtered to obtain the geometric moments of the irregular non-prism target. Among these, the centroid pixel coordinates of the prism target with the center of the fitted circle as the centroid pixel coordinates, the global centroid pixel coordinates of the regular circular target, and the geometric moments of the irregular non-prism target are centroid data.

[0047] In some embodiments, for prism targets with high reflectivity, high-precision centroid extraction is achieved through "adaptive threshold segmentation—morphological optimization—minimum circumcircle fitting": a. Adaptive threshold segmentation: Based on the grayscale difference between the prism reflection area and the background, an adaptive threshold algorithm is used to automatically determine the segmentation threshold, retaining the high-reflectivity area (prism) as the foreground and removing the background area to generate a binarized image, achieving preliminary separation between the reflection area and the background; b. Morphological noise suppression: A 5×5 rectangular structuring element is constructed, and an opening operation is performed on the above binarized image, i.e., erosion followed by dilation, to remove small noise points around the reflection area and optimize the integrity of the reflection area contour; c. Minimum circumcircle fitting localization: For the optimized reflection area contour, the minimum circumcircle fitting algorithm is executed, and the center of the fitted circle is used as the centroid pixel coordinates of the prism target.

[0048] In some embodiments, for a regular circular target consisting of a black substrate and a central white dot, a hierarchical process of "multi-threshold segmentation—contour filtering—precise center localization" is adopted: a. Black substrate region segmentation and optimization: Multi-threshold traversal segmentation: Using a reverse threshold segmentation method, the initial grayscale threshold is set to 50, and traversal segmentation is performed with a step size of 20 increasing to 250. The black substrate region is retained as the foreground (white), and the background region is set to black, generating a binary image of the substrate to ensure complete extraction of the substrate region; Morphological operations: Constructing a 5×5 elliptical structural element, and sequentially performing closing operations (filling holes inside the substrate) and opening operations (removing background noise) to optimize the continuity of the substrate contour; b. Circularity filtering of substrate contour: Contour extraction: Extracting the outermost contour of the binary image of the substrate using a contour detection algorithm, retaining only the outer boundary contour, and eliminating interference from the internal detail contour; Circularity filtering: Calculating the circularity of each contour. : s is the contour area, D is the contour perimeter, and regular contours with a circularity ≥ 0.7 are retained; c. Localization and optimization of the white center: Region of Interest (ROI) clipping: The smallest bounding rectangle that can completely enclose the substrate contour is determined through geometric operations, and this rectangular area is clipped from the preprocessed grayscale image as the ROI for center extraction; Circular detection and localization: Threshold segmentation is performed on the ROI region to extract the white center, and the region that conforms to the circular characteristics is located using a circular detection algorithm, and the pixel coordinates and radius of the center are initially determined; Fitting compensation: If the circular detection fails due to partial occlusion of the center, least squares circle fitting is performed on the white center contour within the ROI, and the model is as follows. Solve for the local coordinates of the circle's center; sub-pixel level precise center localization: perform fitting on the grayscale distribution of 5×5 pixels around the white circle's center, the model is... The gray-level extreme points are calculated to obtain the final global centroid pixel coordinates (u0, v0). Here, (x, y) represents the pixel coordinates within a 5×5 region; G(x, y) represents the gray-level value at the corresponding pixel location; and A represents the gray-level amplitude, the difference between the peak value and the background gray-level, reflecting the gray-level contrast of the central region. The standard deviation of the grayscale distribution. The standard deviation of the distribution in the horizontal direction describes the degree of diffusion of gray values ​​in the central region in the horizontal direction. The vertical standard deviation describes the degree of diffusion of gray values ​​in the central region in the vertical direction.

[0049] In some embodiments, for irregular non-prismatic targets in enhanced image data, an edge detection algorithm is used to extract the edge contour of the target by setting a dual threshold range (low threshold 50-80, high threshold 150-200), balancing weak edge preservation and false edge filtering to ensure the complete extraction of the target's main edge. Contour length filtering: After extracting all edge contours, filter out short contours with a length <50 pixels, as these contours are mostly background noise; retain the target subject contour. Moment invariant centroid calculation: For the filtered target contours, calculate their geometric moments (including the zeroth moment m). 00 First-order moment m 10 and m 01 The formula for calculating (u0, v0) is: .

[0050] S4: Based on the centroid data, the pixel angle coordinates are transformed using the calibration model to obtain the coordinate transformation result.

[0051] In some embodiments, the processor can input centroid data into the calibration model to obtain the corresponding calculated value, and use the conversion residual control to compare with the calibration point reference value to ensure that the initial angle (α1, β1) has a precision control residual ≤ ±0.3″.

[0052] S5: Based on the coordinate transformation results, perform multi-dimensional error correction to obtain the correction results.

[0053] In some embodiments, the processor can perform systematic error correction, environmental error correction, and random error correction based on the coordinate transformation results, respectively, to obtain systematic error correction results, environmental error correction results, and random error correction results; wherein, the systematic error correction results, environmental error correction results, and random error correction results are correction results.

[0054] The systematic error correction is obtained from the total station's factory calibration report: the systematic error is the horizontal angle Δα. sys Elevation angle Δβ sys .

[0055] Environmental error correction is based on temperature T (°C) and air pressure P (hPa) collected by the environmental sensing module, and the impact of environmental factors on the measurement is calculated.

[0056] Random error correction involves continuously collecting 10 sets of data (u0, v0) and calculating 10 sets of angles (α1-α3, β1-β3). Outliers are removed using the 3σ criterion, and the remaining data are averaged to obtain the final angles: α_final = average value (α3), β_final = average value (β3).

[0057] In some embodiments, the expression for the system error correction result is: ; ; The expression for the environmental error correction result is: ; ; ; ; in, This indicates the horizontal angle after only systematic error correction. This represents the horizontal angle observation value before correction. Indicates the systematic error of the horizontal angle. This indicates the elevation angle after only systematic error correction. This represents the elevation angle observation value before correction. Indicates the systematic error of the elevation angle. This represents the environmental error amount of the horizontal angle. Indicates the ambient temperature. Indicates atmospheric pressure. This represents the final horizontal angle after system and environmental error correction. This represents the environmental error amount of the horizontal angle. This represents the final elevation angle after system and environmental error correction. This represents the environmental error amount of the elevation angle.

[0058] S6: Perform edge and corner measurement and accuracy verification on the correction results to obtain the visual edge and corner measurement results, and complete the visual edge and corner measurement of the image total station.

[0059] The visual angle measurement results are the final total station visual angle measurement results, including complete angle data: horizontal angle, elevation angle, horizontal distance, and vertical distance; where α_final = horizontal angle, β_final = elevation angle, Dh = horizontal distance, and Dv = vertical distance.

[0060] In some embodiments, the expressions for horizontal and vertical distances are: ; ; in, Indicates horizontal distance. Indicates the distance to the target. Indicates the elevation angle. Indicates vertical distance.

[0061] In some embodiments, the comparison between the results of this method and manual measurement is shown in Table 1.

[0062] Table 1 shows the difference between the patented method and the results of manual measurement (the difference between the three levels of measurement is 3″).

[0063] In some embodiments, the processor can use Hough transform to initially locate the circle, then extract the outline of the circle, and use RANSAC circle fitting for refinement (because after Hough transform coarse positioning, the extracted outline will inevitably have noise / interference points, and ordinary least squares will amplify these deviations, while RANSAC can correct the deviations). The process includes: First step: Hough transform coarse positioning to obtain the initial center coordinates and radius; Second step: Target outline based on coarse positioning map; Third step: RANSAC circle fitting; Fourth step: If the proportion of effective interior points in RANSAC circle fitting is lower than a set threshold (70%), then the coarse positioning centroid of Hough transform is used as the output to ensure the usability of the results; Among them, point 2 was observed under poor light conditions and windy evening conditions, which also meets the requirements of third-order measurement.

[0064] The advantages are: removal of residual outliers: even if there are still a few noise points in the contour (such as light reflection, edge break points), RANSAC can automatically filter them and fit only the effective interior points; at the same time, it can correct Hough bias: the parameters of the Hough transform are affected by the voting mechanism and preprocessing, and may have slight biases. RANSAC corrects them based on the geometrically optimal solution, making the center and radius closer to the true values.

[0065] In some embodiments of this specification, a total station visual corner measurement method with elevation angle adaptation calibration is used. This method replaces traditional ATR (Automatic Target Recognition) laser aiming with visual technology. It combines low / high elevation angle segmented calibration model optimization to improve the pixel-angle conversion accuracy of the total station. It also integrates Hough transform-center fitting extraction of regular circular targets with manual initial calibration-automatic retest closed-loop logic to achieve fully automatic corner measurement with third-order accuracy in short-to-medium distance engineering scenarios.

[0066] (1) Significantly improved adaptability to complex environments: The pure vision measurement scheme replaces the traditional ATR laser aiming, effectively solving the measurement failure problem in strong light and occlusion scenarios, while achieving accurate measurement of non-prism targets, greatly expanding the range of environmental adaptability; (2) The accuracy of full elevation angle measurement has been greatly optimized: By using the low / high elevation angle segmented calibration model, the influence of elevation angle on coordinate transformation is significantly reduced, the measurement accuracy is stable and meets the requirements of high-precision engineering specifications, and the data reliability is significantly enhanced; (3) The degree of automation and processing efficiency of measurement are significantly improved: After manual initial calibration, no intervention is required throughout the process, realizing a fully automatic edge and corner measurement process. At the same time, the useless data processing is reduced by predicting the target pixel range, and the overall measurement and data processing efficiency is greatly improved. (4) The target compatibility and scene adaptability are greatly enhanced: it can be adapted to various target types such as prism, regular circle, and irregular non-prism, and is widely applicable to engineering surveying, cultural relic mapping, deformation monitoring and other fields, and its practical value is significantly improved.

Claims

1. A visual corner measurement method of an image total station with height angle adaptive calibration, characterized in that, The method comprises the following steps: S1: obtaining the initial height angle in the point position library of the image total station, matching the angles, and measuring by using the calibration model to obtain the visual edge angle collection image; S2: preprocessing the visual edge angle collection image to obtain enhanced image data; S3: extracting the centroid of the enhanced image data of multiple types to obtain centroid data; S4: converting the pixel angle coordinates based on the centroid data by using the calibration model to obtain the coordinate conversion result; S5: correcting the multi-dimensional errors based on the coordinate conversion result to obtain the correction result; S6: measuring the edge angle and verifying the accuracy of the correction result to obtain the visual edge angle measurement result, and completing the visual edge angle measurement of the image total station.

2. The vision angle measurement method of the height angle adaptive calibration total station according to claim 1, wherein, The S1 comprises: obtaining the initial height angle in the point position library of the image total station, enabling the low-height-angle model when the height angle is less than or equal to 30°, enabling the high-height-angle model when the height angle is greater than 30°, using a 12mm focal length when the distance is less than 300m, using a 200mm focal length when the distance is between 300m and 600m, and performing measurement under the condition that the exposure time is not higher than 1 / 500s in strong light and not higher than 1 / 10s in weak light to obtain the visual edge angle collection image; wherein the low-height-angle model and the high-height-angle model belong to the calibration model.

3. The vision angle measurement method of the height angle adaptive calibration total station according to claim 2, characterized in that, The calibration model further comprises a pixel-angle nonlinear relationship, the expression of the low-height-angle model is: ； the expression of the high-height-angle model is: ； the expression of the pixel-angle nonlinear relationship is: ； wherein represents a low elevation angle proportional compensation coefficient, represents a high elevation angle proportional compensation coefficient, represents a horizontal angle conversion coefficient, represents a horizontal pixel, represents a vertical pixel, represents a real-time elevation angle, represents an elevation angle conversion coefficient.​​​​​​​​​​​ 4. The vision angle measurement method of the height angle adaptive calibration total station according to claim 1, wherein, The S2 comprises: collecting the visual edge angle collection image in the region to obtain the multi-region visual edge angle collection image; using a filter kernel to jointly denoise the multi-region visual edge angle collection image by Gaussian filtering and median filtering to obtain the denoising result; performing gray scale and contrast enhancement processing on the denoising result to obtain the enhanced image data.

5. The vision angle measurement method of the height angle adaptive calibration total station according to claim 1, wherein, The S3 comprises: for the prism target with high reflection characteristics in the enhanced image data, sequentially performing processing by using adaptive threshold segmentation, morphological noise suppression, and minimum circumscribed circle fitting positioning to obtain the centroid pixel coordinates of the prism target as the center of the fitted circle; for the regular circular target in the enhanced image data, sequentially performing processing by using black substrate region segmentation and optimization, circularity screening of the substrate contour, and positioning and optimization of the white center to obtain the global centroid pixel coordinates of the regular circular target; for the irregular non-prism target in the enhanced image data, using an edge detection algorithm, setting a double-threshold range to extract the edge contour of the target, and screening the contour length to obtain the geometric moment of the irregular non-prism target; wherein the centroid pixel coordinates of the prism target, the global centroid pixel coordinates of the regular circular target, and the geometric moment of the irregular non-prism target belong to the centroid data.

6. The vision angle measurement method of the height angle adaptive calibration total station according to claim 1, wherein, The S5 comprises: based on the coordinate conversion result, respectively performing system error correction, environmental error correction, and random error correction to respectively obtain the system error correction result, the environmental error correction result, and the random error correction result; wherein the system error correction result, the environmental error correction result, and the random error correction result belong to the correction result.

7. The vision angle measurement method of the height angle adaptive calibration total station according to claim 6, characterized in that, the expression of the system error correction result is: ； ； An expression of the environment error correction result is: ； ； ； ； wherein, denotes the horizontal angle after system error correction only, denotes the horizontal angle observation before correction, denotes the horizontal angle system error, denotes the altitude angle after system error correction only, denotes the altitude angle observation before correction, denotes the altitude angle system error, denotes the amount of environmental error for the horizontal angle, denotes the temperature of the environment, denotes the atmospheric pressure, denotes the final horizontal angle after system + environmental error correction, denotes the amount of environmental error for the horizontal angle, denotes the final altitude angle after system + environmental error correction, denotes the amount of environmental error for the altitude angle.

8. The vision angle measurement method of the height angle adaptive calibration total station according to claim 1, wherein, The visual edge angle measurement result includes a horizontal distance and a vertical distance, and expressions of the horizontal distance and the vertical distance are: ； ； wherein, denotes the horizontal distance, denotes the target distance, denotes the height angle, denotes the vertical distance.

9. A height angle adaptive collimation image total station visual corner measurement system for performing the height angle adaptive collimation image total station visual corner measurement method according to any one of claims 1-8, characterized in that, Comprise: An engineering survey main body comprising a total station host: a horizontal angle encoder, a height angle encoder, a ranging module, and an optical path system, for obtaining an initial height angle in a point position library of an image total station; An embedded imaging unit comprising a long-focus CMOS module and an atmospheric scattering filter, for obtaining a visual edge angle collection image by angle matching and using a calibration model for measurement; An automatic control unit comprising a processor and a special image processing chip, for controlling focal length switching, imaging parameters, measurement process triggering, and calibration model matching; A data processing unit comprising a height angle adaptive calibration model, a multi-target extraction algorithm, target pixel range prediction logic, and an engineering target point position library, for preprocessing the visual edge angle collection image to obtain enhanced image data; Multi-type target centroid extraction is performed on the enhanced image data to obtain centroid data; Based on the centroid data, pixel angle coordinate conversion is performed using the calibration model to obtain a coordinate conversion result; Edge angle measurement and precision verification are performed on the correction result to obtain a visual edge angle measurement result, and the visual edge angle measurement of the image total station is completed; An environment sensing module comprising a temperature sensor, a barometric pressure sensor, and a height angle collection module, for collecting environment parameters to provide a basis for error correction and calibration model calling.

Images