Rapid fusion method and device for steel rail profile data

Abstract

The invention provides a rapid fusion method and device for steel rail profile data. The method comprises the following steps: acquiring a plurality of steel rail profile images acquired from multiple view angles; calculating a first weight of each steel rail profile image based on pixel intensity; calculating a second weight of each steel rail profile image based on a global gradient; determining a weight item of each steel rail profile image according to the first weight and the second weight of each steel rail profile image; according to the weight item of each steel rail profile image, pyramid fusion is carried out on each steel rail profile image, and a fused image of each steel rail profile image is obtained; according to the invention, the shape prior of the steel profile can be fully utilized, and the speed and quality of steel rail profile image fusion are improved.

Description

Technical Field

[0001] The present invention relates to the field of track detection technology, and in particular to a method and device for quickly fusing rail profile data. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention recited in the claims. No admission is made that the description herein is prior art by inclusion in this section.

[0003] At present, special rail inspection vehicles use laser 3D camera technology to quickly survey the rail profile. Since many irregular reflective mirrors are formed on the rail surface during driving, laser imaging is very sensitive to light. Many light spots or dark spots caused by mirror reflection will inevitably appear in the formed light strip, which will eventually seriously affect the system detection results.

[0004] The currently proposed multi-exposure image fusion algorithms include spatial consistency fusion method, gradient descent method, GEF (multi-scale gradient) fusion algorithm and MEFAW (multi-exposure image) fusion algorithm, etc. However, these methods do not meet the engineering requirements of track profile image fusion and have disadvantages such as slow calculation speed and poor imaging result quality. Summary of the invention

[0005] The embodiment of the present invention provides a method for rapid fusion of rail profile data, which is used to effectively solve the bottlenecks of the prior art in terms of fusion quality and speed; the method for rapid fusion of rail profile data comprises:

[0006] Acquire multiple rail profile images collected from multiple perspectives;

[0007] Calculate a first weight based on pixel intensity for each rail profile image;

[0008] Calculate a second weight based on the global gradient for each rail profile image;

[0009] Determining a weight item for each rail profile image according to the first weight and the second weight of each rail profile image;

[0010] According to the weight item of each rail profile image, pyramid fusion is performed on each rail profile image to obtain a fused image of each rail profile image.

[0011] The embodiment of the present invention further provides a fast fusion device for rail profile data, which is used to effectively solve the bottlenecks of the prior art in terms of fusion quality and speed; the fast fusion device for rail profile data comprises:

[0012] A rail profile image acquisition module is used to acquire multiple rail profile images collected from multiple viewing angles;

[0013] A first weight calculation module, used for calculating a first weight of each rail profile image based on pixel intensity;

[0014] A second weight calculation module, used for calculating a second weight of each rail profile image based on a global gradient;

[0015] A weight item determination module, used to determine the weight item of each rail profile image according to the first weight and the second weight of each rail profile image;

[0016] The fused image acquisition module is used to perform pyramid fusion on each rail profile image according to the weight item of each rail profile image to obtain a fused image of each rail profile image.

[0017] An embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for rapid fusion of rail profile data when executing the computer program.

[0018] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for rapidly fusing rail profile data described above is implemented.

[0019] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the method for rapidly fusing rail profile data is implemented.

[0020] The method and device for rapid fusion of rail profile data of the embodiment of the present invention obtain multiple rail profile images collected from multiple perspectives; calculate the first weight of each rail profile image based on pixel intensity; calculate the second weight of each rail profile image based on the global gradient; determine the weight item of each rail profile image according to the first weight and the second weight of each rail profile image; perform pyramid fusion on each rail profile image according to the weight item of each rail profile image to obtain a fused image of each rail profile image; the embodiment of the present invention makes full use of the shape prior of the rail profile to improve the speed and quality of rail profile image fusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0022] Figure 1 This is an example diagram of the effect of mirror emission on laser imaging in an embodiment of the present invention;

[0023] Figure 2 This is an example diagram of a method for rapid fusion of rail profile data in an embodiment of the present invention;

[0024] Figure 3 A specific example diagram of a rail profile image in an embodiment of the present invention;

[0025] Figure 4 is a cumulative histogram of a multiple exposure image sequence in an embodiment of the present invention;

[0026] Figure 5 is a grayscale statistical diagram of a multiple exposure image sequence in an embodiment of the present invention;

[0027] Figure 6 A specific example diagram of a fused image of a rail profile image obtained in an embodiment of the present invention;

[0028] Figure 7 A specific example diagram of using the Roberts operator to perform edge detection on the fused image of each rail profile image in an embodiment of the present invention;

[0029] Figure 8 This is a structural example diagram of a device for rapid fusion of rail profile data in an embodiment of the present invention;

[0030] Fig. 9 The figure is a specific example of the structure of the rapid fusion device of rail profile data in the embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Figure 1 : is an example diagram of the effect of mirror emission on laser imaging in an embodiment of the present invention. Figure 1 As shown, the inventors found that due to the formation of many irregular reflective mirrors on the rail surface during driving, laser imaging is very sensitive to light, and many light spots or dark spots caused by mirror reflection will inevitably appear in the formed light strip, which will eventually seriously affect the system detection results.

[0033] In view of the problems existing in the prior art, the inventor proposes a method of using multiple rail profile images collected from multiple perspectives and fusing the rail profile images.

[0034] Figure 2 FIG. 1 is an example diagram of a method for rapidly fusing rail profile data in an embodiment of the present invention. Figure 2 As shown, the method includes:

[0035] Step 201, obtaining a plurality of rail profile images collected from multiple perspectives;

[0036] Step 202, calculating a first weight of each rail profile image based on pixel intensity;

[0037] Step 203, calculating a second weight of each rail profile image based on the global gradient;

[0038] Step 204, determining a weight item of each rail profile image according to the first weight and the second weight of each rail profile image;

[0039] Step 205: Perform pyramid fusion on each rail profile image according to the weight item of each rail profile image to obtain a fused image of each rail profile image.

[0040] In an embodiment, calculating the first weight of each rail profile image based on pixel intensity may include: calculating the first weight of each rail profile image based on pixel intensity according to pixel intensity of each rail profile image and the mean pixel intensity.

[0041] In specific implementation, the first weight of each rail profile image based on pixel intensity can be calculated according to the following formula:

[0042]

[0043] Among them, W 1,n (x, y) is the first weight of the pixel at position (x, y) of the nth rail profile image based on the pixel intensity, m n is the mean pixel intensity of the nth rail profile image, I n (x, y) is the pixel intensity of the pixel at the position (x, y) of the nth rail profile image, σ n According to m n The gap is used to control the first weight:

[0044]

[0045] α is a constant. In practice, α can be taken as 0.75; N is the number of rail profile images.

[0046] Figure 3 FIG. 1 is a specific example of a rail profile image in an embodiment of the present invention; Figure 3As shown in the figure, affected by the input rail profile images, the pixel intensity of the black background of a large part of the rail profile images is 0, so the average pixel intensity of the four input rail profile images is very similar and a very small value. Therefore, for the four input rail profile images, the first weight based on pixel intensity will give each rail profile image a pixel intensity closer to 1-m. n The pixels of each image are given high weights. n are all very small, so the points assigned high weights are the points with large pixel intensity, that is, the main part of the rail profile, while the edges and noise with small pixel intensity will be given a very low weight, thereby achieving the effect of improving accuracy and filtering noise.

[0047] In an embodiment, calculating the second weight of each rail profile image based on the global gradient may include: calculating the second weight of each rail profile image based on the global gradient according to the gradient of pixel intensity of pixels of each rail profile image.

[0048] Figure 4 , Figure 5 is a cumulative histogram and grayscale statistics diagram of a multiple exposure image sequence in an embodiment of the present invention, such as Figure 4 , Figure 5 As shown in the figure, in the low-exposure image, the pixels in the dark area are saturated to a value close to 0, while the pixel values ​​in the bright area vary greatly. Therefore, the bright area of ​​the low-exposure image will have a higher contrast (large gradient of pixel values), while in the high-exposure image, the opposite relationship will occur.

[0049] Based on this, in specific implementation, the second weight of each rail profile image based on the global gradient can be calculated according to the following formula:

[0050]

[0051] Among them, W 2,n (x, y) is the second weight of the pixel at position (x, y) of the nth rail profile image based on the global gradient, I n (x, y) is the pixel intensity of the pixel at the position (x, y) of the nth rail profile image, Grad n (I n (x,y) -1 is the gradient of the pixel intensity of the pixel at the (x, y) position of the nth rail profile image, and ε is a constant.

[0052] The purpose of the second weight based on the global gradient is to make the well-exposed part of the low-exposure image obtain a high weight, and the overexposed part of the high-exposure image obtain a low weight. The gradient of the low-exposure image is larger when the pixel intensity is smaller, so the part with a small gradient belongs to the bright area. The numerator of W2 is the inverse of the gradient, so the bright area (the part with a good exposure effect) obtains a high weight. The gradient of the high-exposure image is larger when the pixel intensity is larger, so the part with a large gradient belongs to the bright area, so the bright area (the overexposed part) has a small weight, and the dark area closer to the ideal exposure effect obtains a high weight.

[0053] In an embodiment, determining the weight item of each rail profile image according to the first weight and the second weight of each rail profile image may include: multiplying the first weight of each rail profile image by the second weight to obtain the weight item of each rail profile image.

[0054] For example, the weights W1 and W2 obtained from each rail profile image are dot-multiplied to obtain the weight term W.

[0055] Figure 6 FIG. 1 is a specific example of a fused image of a rail profile image obtained in an embodiment of the present invention. Figure 6 As shown, according to the weight item of each rail profile image, each rail profile image is pyramid fused to obtain a fused image of each rail profile image, which may include:

[0056] Step 601: for each rail profile image, repeatedly perform the following steps according to the preset number of pyramid layers: downsampling the weight item once;

[0057] Step 602: Each time the downsampling result is stored in an array to obtain a Gaussian pyramid;

[0058] Step 603: for each rail profile image, repeatedly perform the following steps according to the preset number of pyramid layers: downsampling the rail profile image once and then upsampling it;

[0059] Step 604, each time storing the difference between the rail profile image before downsampling and the result of downsampling and upsampling into an array to obtain a Laplacian pyramid;

[0060] Step 605: perform corresponding point multiplication on each layer of the Gaussian pyramid and each layer of the Laplacian pyramid to obtain a fused pyramid;

[0061] Step 606: traverse the fusion pyramid starting from the highest layer of the fusion pyramid, and take the sum of the upsampling of each layer and the next layer as the fusion result of each layer, so as to obtain a fused image of each rail profile image.

[0062] For example, the number of pyramid layers is set to 7, and the weight item of each rail profile image is downsampled once, and the result of each downsampling is stored in an array, and the above result is repeated 7 times for the downsampling result to obtain a Gaussian pyramid;

[0063] Still setting the number of pyramid layers to 7, downsample each input rail profile image once, then store the difference between the result before downsampling and the result after downsampling and upsampling into an array, repeat the above results 7 times for the downsampled result to obtain the Laplacian pyramid;

[0064] Finally, we multiply each layer of the two pyramids to get the fused pyramid. Starting from the highest layer of the pyramid, we traverse the entire fused pyramid to get the final fused image according to the fusion result equal to the sum of the upsampling of the next layer and the current layer result.

[0065] Among them, since the input rail profile image is a plurality of rail profile images collected from multiple perspectives, in a specific implementation, the final fused image obtained by traversing the entire fusion pyramid can be a fused image of the plurality of rail profile images collected from multiple perspectives.

[0066] One of the biggest advantages of pyramid fusion is that it can achieve very smooth and natural transitions. By fusing at different frequency layers, the low-frequency part controls the large-scale features of the image (such as background, color, etc.), while the high-frequency part retains the details and edge information in the image. In this way, the fused image is visually very smooth, without abrupt edges or seams. At the same time, the result of pyramid fusion has the effect of avoiding seams, which helps to remove noise in the next step of edge detection.

[0067] In the embodiment, after each rail profile image is fused, the method may further include: using a cross differential algorithm to perform edge detection on the fused image of each rail profile image, and removing clustering areas with pixel blocks less than a threshold.

[0068] For example, the Roberts operator is used to detect the edge of each fused image of the rail profile image: the Roberts operator is also called the cross differential algorithm, which is a gradient algorithm based on cross differential, and detects edge lines through local differential calculation. It is often used to process steep low-noise images. When the image edge is close to positive 45 degrees or negative 45 degrees, the algorithm has a better processing effect. Its disadvantage is that the edge positioning is not very accurate, and the extracted edge lines are thicker.

[0069] The template of the Roberts operator is divided into horizontal and vertical directions, as shown in the following formula. It can be seen from its template that the Roberts operator can better enhance the image edges of positive and negative 45 degrees.

[0070]

[0071] Among them, for a pixel with coordinates (x, y) in the image, d x Indicates the difference in grayscale value with adjacent pixels in the horizontal direction (x-axis direction); d y Represents the difference in grayscale value with adjacent pixels in the vertical direction (y-axis direction).

[0072] Figure 7 FIG. 1 is a specific example of using the Roberts operator to perform edge detection on the fused image of each rail profile image in an embodiment of the present invention, as shown in FIG. Figure 7 As shown in the figure, (a) is the edge detection result of Robert operator; (b) is the result after removing the clustering area with pixel blocks less than the threshold; (c) is the final fusion result; a clearer rail profile image can be obtained through the Roberts operator.

[0073] As described above, a fast fusion method for rail profile data proposed in an embodiment of the present invention can effectively solve the problem of data loss caused by mirror emission in vehicle-mounted dynamic detection of rail profile; wherein, the embodiment of the present invention is based on a multi-exposure fusion algorithm based on pyramid fusion combined with edge detection, which makes full use of the shape prior of the steel profile and improves the speed and quality of profile image fusion; and, the method for fusing the weight term obtained by the first weight and the second weight designed in an embodiment of the present invention can achieve smoothing of the fusion result and further improve the fusion quality.

[0074] The present invention also provides a fast fusion device for rail profile data, as described in the following embodiments. Since the principle of the device to solve the problem is similar to the fast fusion method for rail profile data, the implementation of the device can refer to the implementation of the fast fusion method for rail profile data, and the repeated parts will not be repeated.

[0075] Figure 8 FIG. 1 is a structural example diagram of a device for rapidly fusing rail profile data in an embodiment of the present invention. Figure 8 As shown, the device comprises:

[0076] A rail profile image acquisition module 801 is used to acquire multiple rail profile images acquired from multiple viewing angles;

[0077] A first weight calculation module 802, used to calculate a first weight of each rail profile image based on pixel intensity;

[0078] A second weight calculation module 803, used to calculate a second weight of each rail profile image based on a global gradient;

[0079] A weight item determination module 804, configured to determine a weight item for each rail profile image according to the first weight and the second weight of each rail profile image;

[0080] The fused image acquisition module 805 is used to perform pyramid fusion on each rail profile image according to the weight item of each rail profile image to obtain a fused image of each rail profile image.

[0081] In one embodiment, the first weight calculation module 802 is specifically used to calculate the first weight based on pixel intensity of each rail profile image according to the pixel intensity of the pixel points of each rail profile image and the average pixel intensity.

[0082] In one embodiment, the first weight calculation module 802 is specifically used to calculate the first weight of each rail profile image based on pixel intensity according to the following formula:

[0083]

[0084] Among them, W 1,n (x, y) is the first weight of the pixel at position (x, y) of the nth rail profile image based on the pixel intensity, m n is the mean pixel intensity of the nth rail profile image, I n (x, y) is the pixel intensity of the pixel at the position (x, y) of the nth rail profile image, σ n According to m n The gap is used to control the first weight:

[0085]

[0086] α is a constant, and N is the number of rail profile images.

[0087] In one embodiment, the second weight calculation module 803 is specifically configured to calculate the second weight of each rail profile image based on the global gradient according to the gradient of pixel intensity of the pixel points of each rail profile image.

[0088] In one embodiment, the second weight calculation module 803 is specifically used to calculate the second weight of each rail profile image based on the global gradient according to the following formula:

[0089]

[0090] Among them, W 2,n (x, y) is the second weight of the pixel at position (x, y) of the nth rail profile image based on the global gradient, I n (x, y) is the pixel intensity of the pixel at the position (x, y) of the nth rail profile image, Grad n (In (x,y) -1 is the gradient of the pixel intensity of the pixel at the (x, y) position of the nth rail profile image, and ε is a constant.

[0091] In one embodiment, the weight item determination module 804 is specifically configured to multiply the first weight of each rail profile image by the second weight to obtain the weight item of each rail profile image.

[0092] In one embodiment, the fused image acquisition module 805 is specifically used to repeatedly perform the following steps for each rail profile image according to a preset number of pyramid layers: downsampling the weight item once; storing the downsampling result into an array each time to obtain a Gaussian pyramid; repeatedly performing the following steps for each rail profile image according to a preset number of pyramid layers: downsampling the rail profile image once and then upsampling it; storing the difference between the rail profile image before downsampling and the result of downsampling and upsampling into an array each time to obtain a Laplacian pyramid; performing corresponding point multiplication on each layer of the Gaussian pyramid and each layer of the Laplacian pyramid to obtain a fused pyramid; traversing the fused pyramid starting from the highest layer of the fused pyramid, taking the sum of the upsampling of each layer and the next layer as the fusion result of each layer, and obtaining a fused image of each rail profile image.

[0093] Fig. 9 FIG. 1 is a specific example diagram of the structure of the rapid fusion device for rail profile data in an embodiment of the present invention. Fig. 9 As shown, in one embodiment, Figure 8 The device for rapid fusion of rail profile data in the embodiment of the present invention may further include an edge detection module 901 .

[0094] In this example, the edge detection module 901 is specifically used to use a cross differential algorithm to perform edge detection on the fused image of each rail profile image, and remove clustering areas with pixel blocks less than a threshold.

[0095] An embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for rapid fusion of rail profile data when executing the computer program.

[0096] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for rapidly fusing rail profile data described above is implemented.

[0097] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the method for rapidly fusing rail profile data is implemented.

[0098] The method and device for rapid fusion of rail profile data of the embodiment of the present invention obtain multiple rail profile images collected from multiple perspectives; calculate the first weight of each rail profile image based on pixel intensity; calculate the second weight of each rail profile image based on the global gradient; determine the weight item of each rail profile image according to the first weight and the second weight of each rail profile image; perform pyramid fusion on each rail profile image according to the weight item of each rail profile image to obtain a fused image of each rail profile image; the embodiment of the present invention makes full use of the shape prior of the rail profile to improve the speed and quality of rail profile image fusion.

[0099] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0100] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0101] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0102] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0103] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for rapid fusion of rail profile data, characterized in that: include: Acquire multiple rail profile images collected from multiple perspectives; Calculate a first weight based on pixel intensity for each rail profile image; Calculate a second weight based on the global gradient for each rail profile image; Determining a weight item for each rail profile image according to the first weight and the second weight of each rail profile image; According to the weight item of each rail profile image, pyramid fusion is performed on each rail profile image to obtain a fused image of each rail profile image.

2. The method according to claim 1, characterized in that Calculate the first weight of each rail profile image based on pixel intensity, including: According to the pixel intensity of the pixels of each rail profile image and the average pixel intensity, a first weight based on the pixel intensity of each rail profile image is calculated.

3. The method according to claim 2, characterized in that The first weight of each rail profile image based on pixel intensity is calculated according to the following formula: Among them, W 1,n (x, y) is the first weight of the pixel at position (x, y) of the nth rail profile image based on the pixel intensity, m n is the mean pixel intensity of the nth rail profile image, I n (x, y) is the pixel intensity of the pixel at the position (x, y) of the nth rail profile image, σ n According to m n The gap is used to control the first weight: α is a constant, and N is the number of rail profile images.

4. The method according to claim 1, characterized in that Calculate the second weight of each rail profile image based on the global gradient, including: According to the gradient of pixel intensity of each rail profile image, a second weight based on the global gradient is calculated for each rail profile image.

5. The method according to claim 4, characterized in that The second weight of each rail profile image based on the global gradient is calculated according to the following formula: Among them, W 2,n (x, y) is the second weight of the pixel at position (x, y) of the nth rail profile image based on the global gradient, I n (x, y) is the pixel intensity of the pixel at the position (x, y) of the nth rail profile image, Grad n (I n (x,y) -1 is the gradient of the pixel intensity of the pixel at the (x, y) position of the nth rail profile image, and ε is a constant.

6. The method according to claim 1, characterized in that Determining a weight item of each rail profile image according to the first weight and the second weight of each rail profile image includes: The first weight of each rail profile image is multiplied by the second weight to obtain a weight item of each rail profile image.

7. The method according to claim 1, characterized in that According to the weight item of each rail profile image, pyramid fusion is performed on each rail profile image to obtain a fused image of each rail profile image, including: For each rail profile image, the following steps are repeated according to the preset number of pyramid levels: downsampling the weight item once; Each time, the downsampling result is stored in an array to obtain a Gaussian pyramid; For each rail profile image, the following steps are repeated according to the preset number of pyramid layers: downsampling and upsampling the rail profile image once; Each time, the difference between the rail profile image before downsampling and the result of downsampling and upsampling is stored in an array to obtain a Laplacian pyramid; Perform corresponding point multiplication on each layer of the Gaussian pyramid and each layer of the Laplacian pyramid to obtain a fused pyramid; The fusion pyramid is traversed from the highest layer of the fusion pyramid, and the sum of the upsampling of each layer and the next layer is taken as the fusion result of each layer to obtain the fused image of each rail profile image.

8. The method according to claim 1, characterized in that After the fusion image of each rail profile image, it also includes: The cross differential algorithm is used to perform edge detection on the fused image of each rail profile image, and the clustering areas with pixel blocks less than the threshold are removed.

9. A rapid fusion device for rail profile data, characterized in that: include: A rail profile image acquisition module is used to acquire multiple rail profile images collected from multiple viewing angles; A first weight calculation module, used for calculating a first weight of each rail profile image based on pixel intensity; A second weight calculation module, used for calculating a second weight of each rail profile image based on a global gradient; A weight item determination module, used to determine the weight item of each rail profile image according to the first weight and the second weight of each rail profile image; The fused image acquisition module is used to perform pyramid fusion on each rail profile image according to the weight item of each rail profile image to obtain a fused image of each rail profile image.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 8 is implemented.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

12. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images