A Cross-Coordinate System Road Network Data Fusion Method Based on Hierarchical Matching and Non-rigid Registration

Abstract

This invention proposes a cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration, belonging to the field of road network technology, and solves the problems of computational complexity and poor versatility of existing methods. The method includes: acquiring and preprocessing benchmark and target road network data; estimating global transformation relationships by calibrating feature points, and initially aligning the benchmark road network to the target coordinate system to obtain a global region; constructing a fixed image of the target road network and a moving image of the benchmark road network within the global region; using algorithms such as B-spline free deformation that can model local nonlinear deformations to perform non-rigid registration on the fixed and moving images, optimizing the cost function to obtain pixel-level deformation fields; geometrically correcting the benchmark road network based on the deformation field, transferring the target road network attribute information, and summarizing and outputting a fused dataset. This method improves local matching accuracy and versatility through a hierarchical strategy and pixel-level registration, overcoming the influence of differences in road segmentation.

Description

Technical Field

[0001] This invention belongs to the field of road network technology, specifically relating to a cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration. Background Technology

[0002] Road network data from different sources are difficult to directly fuse and transfer attributes due to coordinate system differences (such as nonlinear local offsets between WGS-84 and GCJ-02) and inconsistent road segment segmentation. Existing methods for obtaining road network data mapping relationships, such as vector matching based on geometric features or methods based on global transformation models, have significant limitations in handling such problems. Standard coordinate transformations are difficult to correct complex local nonlinear distortions; traditional vector feature-based matching methods are computationally complex and highly sensitive to differences in segmentation methods such as the start and end points and lengths of road segments when attempting to establish correspondences between different road networks, leading to matching errors or failures and seriously affecting the reliability of attribute transfer. Some methods rely on specific coordinate transformation models, resulting in poor universality. Therefore, there is an urgent need for a new fusion method that can accurately handle local deformations, is efficient and universal, and can robustly establish mapping relationships between road network entities under different segmentation methods. Summary of the Invention

[0003] In view of this, the present invention provides a cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration to solve the problems of computational complexity and poor versatility of existing road network fusion methods.

[0004] The technical solution adopted in this invention is as follows:

[0005] A cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration includes: Step 1: Acquire benchmark road network data and target road network data and perform preprocessing respectively;

[0006] In step 1, the baseline road network data adopts a first coordinate system and includes geometric shapes and unique road segment identifiers, while the target road network data adopts a second coordinate system and is vector data or a geographic reference raster image.

[0007] Step 2: Obtain the global spatial transformation relationship between the baseline road network data and the target road network data, and based on the global spatial transformation relationship, initially align the baseline road network data to the target coordinate system to obtain the overall region;

[0008] Step 2 specifically includes the following steps:

[0009] Step 2.1: Obtain a specified number of corresponding geographic feature points from the baseline road network data and the target road network data, and mark their coordinates to obtain calibration point pairs;

[0010] Step 2.2: Based on the calibration point pairs, use affine or similarity transformation models to estimate the global spatial transformation relationship using the least squares method; in Step 2.2, a spatial weighting mechanism is introduced into the least squares estimation to improve the estimation robustness.

[0011] Step 2.3: Based on the global spatial transformation relationship, the baseline road network data is initially aligned to the target coordinate system to obtain the overall region.

[0012] Step 3: Based on the obtained overall region, construct a fixed image of the target road network and a moving image of the baseline road network;

[0013] Step 3 specifically includes the following steps:

[0014] Step 3.1: Divide the overall area into several smaller blocks and record their geographical boundaries;

[0015] Step 3.2: Extract the image blocks corresponding to several small blocks from the target road network data in Step 3.1, and convert them into grayscale images;

[0016] Step 3.3: Query the reference road network segments that have been initially aligned to the target coordinate system in Step 2, crop out the road segments that intersect with each small geographical area in Step 3.1, convert the coordinates of the cropped road segments into the corresponding image pixel coordinates, and draw them onto a blank canvas of the same size as the fixed image in Step 3.2 to generate a moving image with the same spatial range as the fixed image.

[0017] Step 4: Perform local non-rigid image registration between the fixed image and the moving image. Use an algorithm that can model local nonlinear deformation and obtain the pixel-level deformation field by optimizing the cost function.

[0018] Step 4 specifically includes the following steps:

[0019] Step 4.1: Select a registration algorithm that can model local nonlinear deformation, embed the moving image into a regular grid composed of control points, and achieve smooth image deformation by moving the control points. The original spatial point coordinates... The new position after deformation satisfies:

[0020]

[0021] In the formula, Indicates the first The B-spline basis function values ​​corresponding to each control point in the u direction. Indicates the first The B-spline basis function values ​​of each control point in the v direction. Let w represent the B-spline basis function value corresponding to the nth B-spline basis function in the w direction. Indicates the location of the control point. Indicates the coordinates of the deformed point. Depend on Local coordinates obtained through normalization;

[0022] Step 4.2: Set the fixed image as... The moving image is The registration objective is to find the optimal deformation field that minimizes the following cost function:

[0023]

[0024] in: Represents the image difference measurement function; This represents a regularization term used to ensure the smoothness of deformation; Indicates the balancing weights;

[0025] The deformation field can be determined by the location of the control points:

[0026]

[0027] in Let be the B-spline basis functions, and let represent the B-spline basis functions of the i-th control point in the x-direction and the j-th control point in the y-direction, respectively. It is the deformation vector at the control point;

[0028] Step 4.3: Record the deformation field after convergence in each local region. .

[0029] Step 5: Perform geometric correction on the reference road network data based on the deformation field, and transfer the attribute information of the target road network data to the reference road network through pixel-level mapping. Summarize the results of all local processing units and output the fused road network dataset.

[0030] Step 5 specifically includes the following steps:

[0031] By applying the deformation field to the local original reference road section, a geometry that is precisely aligned to the target coordinate space is obtained;

[0032] If the target image contains pixel-level attribute information, the attribute transfer from the image to the vector road segment is achieved through the deformation field;

[0033] The registration and fusion results of all local regions are merged, and boundary smoothing is applied to output the complete fused dataset.

[0034] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0035] This invention focuses on solving the challenge of cross-coordinate system road network fusion through local non-rigid image registration. Compared to traditional global transformation or vector matching methods, which struggle to handle local nonlinear distortions, suffer from low computational efficiency, and are sensitive to road segmentation, this invention directly captures spatial correspondences at the pixel level. This not only achieves sub-pixel-level precise positioning and effectively compensates for complex deformations, but also features a core registration algorithm suitable for GPU parallel acceleration, demonstrating high efficiency potential. Furthermore, this method does not rely on a specific coordinate transformation model, is highly versatile, applicable to any rasterizable coordinate system pair, and possesses inherent robustness to differences in road segmentation. By unifying data from different sources into the image domain for processing, it also facilitates heterogeneous data fusion, providing strong support for the reliable transmission of traffic conditions. Attached Figure Description

[0036] The present invention will be described by way of example and with reference to the accompanying drawings, wherein:

[0037] Figure 1 This is a schematic diagram of the process structure involved in an embodiment of the present invention.

[0038] Figure 2 This refers to the target matching region 1 in this embodiment of the invention.

[0039] Figure 3 This is the matching effect of matching the target matching region 1 using latitude and longitude matching (traditional method) according to an embodiment of the present invention.

[0040] Figure 4 This is the matching effect of matching the target matching region 1 using the hierarchical matching method of this embodiment of the invention.

[0041] Figure 5 This refers to target matching region 2 in this embodiment of the invention.

[0042] Figure 6 This is the matching effect of using latitude and longitude matching (traditional method) to match the target matching region 2 in the embodiment of the present invention.

[0043] Figure 7 This is the matching effect of matching the target matching region 2 using the layered matching method of this embodiment of the invention.

[0044] Figure 8 The confidence improvement trend of the hierarchical matching algorithm involved in the embodiments of the present invention compared with the monotonic latitude and longitude matching algorithm is shown.

[0045] Figure 9 This refers to the source network data involved in the embodiments of the present invention.

[0046] Figure 10 This refers to the target road network data involved in the embodiments of the present invention.

[0047] Figure 11 This is a visualization of the matching results of the number of source road network segments on the target road network in the embodiments of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0049] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0050] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other.

[0051] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0052] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" of the second feature includes the first feature directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature includes the first feature directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0053] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other.

[0054] Example

[0055] like Figure 1 As shown in the figure, this embodiment of the invention discloses a cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration, including:

[0056] Step 1: Obtain baseline road network data and target road network data, and preprocess them respectively;

[0057] In step 1, the baseline road network data adopts a first coordinate system and includes geometric shapes and unique road segment identifiers, while the target road network data adopts a second coordinate system and is vector data or a geographic reference raster image.

[0058] Step 2: Obtain the global spatial transformation relationship between the baseline road network data and the target road network data, and based on the global spatial transformation relationship, initially align the baseline road network data to the target coordinate system to obtain the overall region;

[0059] Step 2 specifically includes the following steps:

[0060] Step 2.1: Obtain a specified number of corresponding geographic feature points from the baseline road network data and the target road network data, and mark their coordinates to obtain calibration point pairs;

[0061] Step 2.2: Based on the calibration point pairs, use affine or similarity transformation models to estimate the global spatial transformation relationship using the least squares method; in Step 2.2, a spatial weighting mechanism is introduced into the least squares estimation to improve the estimation robustness.

[0062] The spatial weighting mechanism dynamically adjusts the weights based on the spatial distribution characteristics of the calibration points (such as distance from the center point, local point density, or geometric reliability of feature points), giving priority to point pairs with uniform spatial distribution or high confidence, reducing the interference of noise points or outliers on the global transformation estimation, thereby improving the stability and accuracy of the transformation relationship.

[0063] Step 2.3: Based on the global spatial transformation relationship, the baseline road network data is initially aligned to the target coordinate system to obtain the overall region.

[0064] Step 3: Based on the obtained overall region, construct a fixed image of the target road network and a moving image of the baseline road network;

[0065] Step 3 specifically includes the following steps:

[0066] Step 3.1: Divide the overall area into several smaller blocks and record their geographical boundaries;

[0067] Step 3.2: Extract the image blocks corresponding to several small blocks from the target road network data in Step 3.1, and convert them into grayscale images;

[0068] Step 3.3: Query the reference road network segments that have been initially aligned to the target coordinate system in Step 2, crop out the road segments that intersect with each small geographical area in Step 3.1, convert the coordinates of the cropped road segments into the corresponding image pixel coordinates, and draw them onto a blank canvas of the same size as the fixed image in Step 3.2 to generate a moving image with the same spatial range as the fixed image.

[0069] Step 4: Perform local non-rigid image registration between the fixed image and the moving image. Use an algorithm that can model local nonlinear deformation and obtain the pixel-level deformation field by optimizing the cost function.

[0070] Step 4 specifically includes the following steps:

[0071] Step 4.1: Select a registration algorithm that can model local nonlinear deformation, embed the moving image into a regular grid composed of control points, and achieve smooth image deformation by moving the control points, where the original spatial point coordinates... The new position after deformation satisfies:

[0072]

[0073] In the formula, Indicates the first The B-spline basis function value corresponding to each control point in the u-direction measures the distance of the control point from that control point in the u-direction. A larger value indicates that the control point is closer to the control point and contributes more. Indicates the first The B-spline basis function values ​​of each control point in the v direction. Let w represent the B-spline basis function value corresponding to the nth B-spline basis function in the w direction. Indicates the location of the control point. Indicates the coordinates of the deformed point. Depend on Local coordinates obtained through normalization;

[0074] Step 4.2: Set the fixed image as... The moving image is The registration objective is to find the optimal deformation field that minimizes the following cost function:

[0075]

[0076] in: Represents the image difference measurement function; This represents a regularization term used to ensure the smoothness of deformation; Indicates the balancing weights;

[0077] The deformation field can be determined by the location of the control points:

[0078]

[0079] in Let be the B-spline basis functions, and let represent the B-spline basis functions for the i-th control point in the x-direction and the j-th control point in the y-direction, respectively, with the same meaning as above. It is the deformation vector at the control point;

[0080] Step 4.3: Record the deformation field after convergence in each local region. .

[0081] Step 5: Perform geometric correction on the reference road network data based on the deformation field, and transfer the attribute information of the target road network data to the reference road network through pixel-level mapping. Summarize the results of all local processing units and output the fused road network dataset.

[0082] Step 5 specifically includes the following steps:

[0083] By applying the deformation field to the local original reference road section, a geometry that is precisely aligned to the target coordinate space is obtained;

[0084] If the target image contains pixel-level attribute information, the attribute transfer from the image to the vector road segment is achieved through the deformation field;

[0085] The registration and fusion results of all local regions are merged, and boundary smoothing is applied to output the complete fused dataset.

[0086] Specifically, the matching effect of the hierarchical matching method involved in this embodiment is as follows: Figure 2-8 As shown:

[0087] Figure 2 For target matching region 1, Figure 3 The image shows the matching effect of using latitude and longitude matching (traditional method) to match target region 1. Figure 4 The matching effect of using the hierarchical matching method involved in this embodiment to match the target matching region 1;

[0088] Figure 5 Match region 2 for target. Figure 6 The image shows the matching effect of using latitude and longitude matching (traditional method) to match target region 2. Figure 7 The matching effect of using the layered matching method of this embodiment to match the target matching region 2;

[0089] Combination Figure 8As can be seen, the local non-rigid registration of layered matching eliminates the tiny offsets that latitude and longitude matching cannot eliminate, and the confidence of the matching is also improved by 0.3462. The final deviation is at the pixel level and can be ignored, showing a very significant effect. In other more complex scenarios, the accuracy of layered matching can still be guaranteed.

[0090] In addition, from Figure 9-11 As can be seen from this embodiment, the hierarchical matching method extracts road segment data from the road network in image form and matches it to the target road network, achieving the technical effect of data assimilation between road networks. Among these, Figure 9 The source road network data is in image format, and includes unknown latitude and longitude coordinate systems. Color codes represent the values ​​of road segment attributes. Figure 10 The target road network data is in SHP format, and the road network includes the WGS84 coordinate system. Figure 11 This is a visualization of the matching results of the number of road segments in the source road network on the target road network.

[0091] The circuits, electronic components, and modules involved are all existing technologies, which can be fully implemented by those skilled in the art, and need not be elaborated upon. The scope of protection of this invention does not involve any improvement to the software and methods.

[0092] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0093] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration, characterized in that, include: Step 1: Obtain baseline road network data and target road network data, and preprocess them respectively; Step 2: Obtain the global spatial transformation relationship between the baseline road network data and the target road network data, and based on the global spatial transformation relationship, initially align the baseline road network data to the target coordinate system to obtain the overall region; Step 3: Based on the obtained overall region, construct a fixed image of the target road network and a moving image of the baseline road network; Step 4: Perform local non-rigid image registration between the fixed image and the moving image. Use an algorithm that can model local nonlinear deformation and obtain the pixel-level deformation field by optimizing the cost function. Step 4.1: Select a registration algorithm that can model local nonlinear deformation, embed the moving image into a regular grid composed of control points, and achieve smooth image deformation by moving the control points, where the original spatial point coordinates... The new position after deformation satisfies: ； In the formula, Indicates the first The B-spline basis function values ​​corresponding to each control point in the u direction. Indicates the first The B-spline basis function values ​​of each control point in the v direction. Let w represent the B-spline basis function value corresponding to the nth B-spline basis function in the w direction. Indicates the location of the control point. Indicates the coordinates of the deformed point. Depend on Local coordinates obtained through normalization; Step 4.2: Set the fixed image as... The moving image is The registration objective is to find the optimal deformation field that minimizes the following cost function: ； in: Represents the image difference measurement function; This represents a regularization term used to ensure the smoothness of deformation; Indicates the balancing weights; The deformation field is determined by the location of the control points: ； in Let be the B-spline basis functions, and let represent the B-spline basis functions of the i-th control point in the x-direction and the j-th control point in the y-direction, respectively. It is the deformation vector at the control point; Step 4.3: Record the deformation field after convergence in each local region. ; Step 5: Perform geometric correction on the reference road network data based on the deformation field, and transfer the attribute information of the target road network data to the reference road network through pixel-level mapping. Summarize the results of all local processing units and output the fused road network dataset.

2. The cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration according to claim 1, characterized in that, In step 1, the baseline road network data adopts a first coordinate system and includes geometric shapes and unique road segment identifiers, while the target road network data adopts a second coordinate system and is vector data or a geographic reference raster image.

3. The cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration according to claim 1, characterized in that, Step 2 specifically includes the following steps: Step 2.1: Obtain a specified number of corresponding geographic feature points from the baseline road network data and the target road network data, and mark their coordinates to obtain calibration point pairs; Step 2.2: Based on the calibration point pairs, use affine or similarity transformation models to estimate the global spatial transformation relationship using the least squares method; Step 2.3: Based on the global spatial transformation relationship, the baseline road network data is initially aligned to the target coordinate system to obtain the overall region.

4. The cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration according to claim 3, characterized in that, In step 2.2, a spatial weighting mechanism is introduced into the least squares estimation.

5. The cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration according to claim 1, characterized in that, Step 3 specifically includes the following steps: Step 3.1: Divide the overall area into several smaller blocks and record their geographical boundaries; Step 3.2: Extract the image blocks corresponding to several small blocks from the target road network data in Step 3.1, and convert them into grayscale images; Step 3.3: Query the reference road network segments that have been initially aligned to the target coordinate system in Step 2, crop out the road segments that intersect with each small geographical area in Step 3.1, convert the coordinates of the cropped road segments into the corresponding image pixel coordinates, and draw them onto a blank canvas of the same size as the fixed image in Step 3.2 to generate a moving image with the same spatial range as the fixed image.

6. The cross-coordinate system road network data fusion method based on hierarchical matching and non-rigid registration according to claim 1, characterized in that, Step 5 specifically includes the following steps: By applying the deformation field to the local original reference road section, a geometry that is precisely aligned to the target coordinate space is obtained; If the target image contains pixel-level attribute information, the attribute transfer from the image to the vector road segment is achieved through the deformation field; The registration and fusion results of all local regions are merged, and boundary smoothing is applied to output the complete fused dataset.

Images