A linear guide rail scanning image real-time denoising method and system

Abstract

The present application relates to the technical field of image data processing, in particular to a kind of linear guide rail scanning image real-time denoising method and system, comprising: obtaining linear guide rail scanning image to be processed;Using improved anisotropic diffusion filter algorithm to linear guide rail scanning image to be processed is denoised, to obtain the linear guide rail scanning image after denoising processing.Wherein, improved anisotropic diffusion filter algorithm includes diffusion function, the diffusion function is positively correlated with initial diffusion function, the artifact boundary confidence of each pixel point, and the artifact boundary confidence is characterized as the credibility of each pixel point belongs to artifact.The present application solves the problem that existing algorithm is not enough when facing complex noise.

Description

Technical Field

[0001] This invention relates to the field of image data processing technology. More specifically, this invention relates to a real-time denoising method and system for scanning images using a linear guide rail. Background Technology

[0002] In the fields of industrial automation and precision manufacturing, linear guides, as core linear motion components of high-precision equipment such as CNC machine tools, semiconductor manufacturing equipment, laser cutting machines, and optical inspection platforms, play a crucial role in precise guidance and smooth transmission. Their operating status directly affects the processing accuracy, operational stability, and service life of the equipment. As the manufacturing industry moves towards higher precision and efficiency, the health monitoring and defect detection of linear guides have become increasingly important for ensuring production quality. To achieve real-time monitoring of the guide surface, industrial cameras, line scan cameras, or laser vision sensors are typically used to perform high-speed scanning imaging of the guide surface, thereby obtaining image data that can be used for surface defect identification and quality assessment.

[0003] However, in actual scanning, the linear guide rails, mostly made of metal with a high degree of smoothness, exhibit strong reflective properties that generate large areas of high brightness during imaging. Furthermore, to prevent oxidation and corrosion of the metal surface, the guide rails are typically coated with an oil film, which not only creates irregular bright spots but may also introduce scattering blurring effects. These imaging artifacts and interference signals often have clear boundaries and significant brightness variations, and are not traditional Gaussian noise or salt-and-pepper noise, but rather complex noise with certain structural characteristics. This type of interference not only masks minor scratches, wear, pitting, and other defects on the guide rail surface but also introduces misjudgments in subsequent defect detection, texture analysis, and surface quality evaluation.

[0004] In recent years, some adaptive denoising methods based on anisotropic diffusion filtering algorithms have been able to remove noise while preserving edges. However, their core judgment criterion is the magnitude of the gradient, meaning that areas with high gradients are identified as edges and protected. In linear guideway scanning images, the gradient values ​​of highly reflective or oil film boundaries are extremely large. Existing algorithms may misidentify these artifacts as real edges that should be protected, thus stopping information diffusion. This results in the artifacts being completely preserved, failing to repair the underlying real structural texture, and consequently leading to insufficient adaptability of existing algorithms when facing complex noise. Summary of the Invention

[0005] To address the problem of insufficient adaptability of existing algorithms in the context of the above background art when facing complex noise, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a real-time denoising method for linear guideway scan images, comprising: acquiring a linear guideway scan image to be processed; and performing denoising processing on the linear guideway scan image to be processed using an improved anisotropic diffusion filtering algorithm to obtain a denoised linear guideway scan image; wherein the improved anisotropic diffusion filtering algorithm includes a diffusion function, the diffusion function being positively correlated with an initial diffusion function and the artifact boundary confidence of each pixel, the artifact boundary confidence representing the confidence that each pixel belongs to an artifact.

[0007] The above technical solution introduces a diffusion function that is positively correlated with the confidence level of the artifact boundary in the anisotropic diffusion filter, which enables the diffusion intensity to be adaptively increased in the artifact region. This effectively weakens the boundary features of artifacts such as reflections and oil films and restores the masked true texture. In the non-artifact region, the original detail protection capability is maintained, realizing differentiated processing for different regions. While ensuring the integrity of the guide rail surface details, it significantly improves the noise reduction effect and the targeting and robustness of the processing.

[0008] Furthermore, the diffusion function for, , For pixels The confidence level of the artifact boundary. is the initial diffusion function.

[0009] The above technical solution uses a weighted fusion of artifact boundary confidence and initial diffusion function. This allows regions with high artifact boundary confidence to retain stronger smoothing capabilities during diffusion, thereby weakening artifact contours that would otherwise be protected due to larger gradients and effectively repairing artifact regions. Meanwhile, regions with low confidence rely more on the initial diffusion function to maintain normal structural details, thus suppressing artifacts while ensuring the integrity of real textures and the overall clarity of the image.

[0010] Furthermore, the confidence level of artifact boundaries for, , For pixels Texture similarity, For pixels The degree of mutation, For pixels gradient magnitude, , These represent the minimum and maximum gradient magnitudes of pixels in the scanned image of the linear guide rail to be processed. This is the normalization function.

[0011] The above technical solution integrates texture similarity, abrupt change degree, and normalized gradient magnitude to form a confidence assessment of whether a pixel belongs to the boundary of an artifact. This results in a higher confidence score in areas that simultaneously have significant directional abrupt changes, low texture similarity, and large gradient magnitude. This allows for accurate location of artifacts such as reflections or oil films, providing a reliable basis for subsequent targeted suppression of artifacts and restoration of the underlying true structure and texture, and effectively improving the accuracy and stability of noise reduction processing.

[0012] Furthermore, texture similarity for, , For the natural constant An exponential function with base 0. In pixels Pixels within the set neighborhood of the center grayscale value, In pixels The average grayscale value of all pixels within a neighborhood of the center is set. For the pixel point gradient magnitude, This refers to the maximum gradient magnitude of the pixels within the defined neighborhood.

[0013] The above technical solution measures texture similarity by combining the weighted difference between the gray values ​​of neighboring pixels and the gradient magnitude. This results in regions with similar gray-level distributions and gentle gradient magnitude changes receiving higher similarity scores, while regions with large gray-level differences or significant gradient abrupt changes have lower similarity scores. This allows the solution to highlight regions with significant differences from surrounding textures while preserving texture details with consistent structures. This provides a precise basis for distinguishing between real textures and interference artifacts during subsequent denoising, thereby improving the overall effect of detail preservation and noise suppression.

[0014] Furthermore, the degree of mutation for, , In pixels Target pixels within the set neighborhood of the center gradient direction, For pixels gradient direction, The total number of target pixels within the defined neighborhood range, wherein the target pixels are pixels whose gradient magnitude is greater than a preset threshold.

[0015] The above technical solution calculates the similarity of the gradient direction of the target pixel and the high-gradient pixels in its neighborhood, and determines the region with low directional consistency as the region with high degree of change. This can effectively distinguish between the regular gradient distribution generated by the edge of the real structure and the chaotic directional distribution caused by artifacts such as reflection and oil film, and achieve accurate identification of artifact regions.

[0016] Furthermore, the defined neighborhood range is 3. 3.

[0017] Furthermore, a CCD or CMOS camera is used to acquire scanned images of the linear guide rail to be processed.

[0018] Furthermore, it also includes performing grayscale processing on the scanned image of the linear guide rail to be processed.

[0019] Furthermore, the normalization function is a max-min normalization.

[0020] In a second aspect, the present invention provides a real-time denoising system for linear guide rail scanned images, comprising a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, implement the real-time denoising method for linear guide rail scanned images as described above.

[0021] The beneficial effects of this invention are as follows:

[0022] This invention introduces an artifact boundary confidence level calculated based on texture similarity, abrupt change, and gradient magnitude into anisotropic diffusion filtering, and adaptively fuses it with the initial diffusion function. This allows for dynamic adjustment of diffusion intensity in different regions during denoising, resulting in stronger smooth restoration in artifact areas such as high reflectivity and oil films, while maintaining high fidelity in areas with realistic structural details. It effectively overcomes the problem of traditional methods misjudging artifact boundaries as real edges and protecting them, improving the balance between artifact suppression and detail preservation. Combined with grayscale preprocessing and normalization calculation, the algorithm exhibits higher stability and adaptability under different lighting, reflectivity, and imaging conditions, while also possessing strong real-time processing capabilities, making it suitable for online real-time denoising of high-precision equipment such as linear guides. Attached Figure Description

[0023] Figure 1 This is a schematic flowchart illustrating a real-time noise reduction method for scanning images using a linear guide rail according to an embodiment of the present invention.

[0024] Figure 2 This is a schematic block diagram illustrating the structure of a real-time noise reduction system for scanning images using a linear guide rail, according to an embodiment of the present invention. Detailed Implementation

[0025] An embodiment of a real-time noise reduction method for scanning images of linear guide rails.

[0026] like Figure 1 As shown, a flowchart of a real-time denoising method for scanning images using a linear guide rail according to an embodiment of the present invention includes the following steps:

[0027] S1: Obtain the scanned image of the linear guide rail to be processed.

[0028] In a preferred embodiment, the linear guideway scan image to be processed is first acquired using a high-resolution imaging device, which can be a CCD (charge-coupled device) camera or a CMOS (complementary metal-oxide-semiconductor) camera.

[0029] Specifically, CCD cameras offer high photosensitivity and imaging quality, enabling them to acquire image data with high signal-to-noise ratio and clear detail even in low-light conditions. CMOS cameras, on the other hand, boast advantages such as low power consumption, fast response speed, and ease of integration, providing stable image acquisition capabilities in high-speed scanning scenarios. By selecting the appropriate camera type, not only can different production environments and inspection needs be optimized, but the accuracy and robustness of subsequent image processing can also be effectively improved, thus providing a high-quality raw data foundation for the detection of surface defects in linear guides.

[0030] After acquiring the scanned image of the linear guide rail to be processed, this embodiment further performs grayscale processing on the image, that is, converts the original color image into a single-channel grayscale image. The technical effect of this step is that by removing color information and retaining only grayscale information reflecting brightness changes, it can effectively reduce data dimensionality and storage pressure, reduce the computational complexity of subsequent image processing, and to a certain extent reduce interference caused by changes in ambient lighting. In addition, grayscale processing can highlight the surface texture, scratches, pits, and other structural features of the linear guide rail.

[0031] S2: An improved anisotropic diffusion filtering algorithm is used to denoise the scanned image of the linear guide rail to be processed.

[0032] In a preferred embodiment, the improved anisotropic diffusion filtering algorithm includes a diffusion function, the diffusion function... for, , For pixels The confidence level of the artifact boundary. is the initial diffusion function.

[0033] The confidence level of the artifact boundary for, , For pixels Texture similarity, For pixels The degree of mutation, For pixels gradient magnitude, , These represent the minimum and maximum gradient magnitudes of pixels in the scanned image of the linear guide rail to be processed. This is the normalization function.

[0034] By multiplying texture similarity, abrupt change, and normalized gradient magnitude, a confidence index for artifact boundaries is constructed that comprehensively reflects the multidimensional features of pixels. Texture similarity reflects the consistency of grayscale distribution in a local area, abrupt change reflects the magnitude of variation in structural direction, and normalized gradient magnitude measures the significance of brightness change at that location. With this design, when the texture similarity of a pixel's region is high, the gradient direction change is significant, and the brightness change is close to the highest level in the entire image, the confidence value will increase significantly, highlighting pixels that are highly likely to be artifact boundaries. Conversely, in areas with low texture similarity, insignificant directional abrupt changes, and weak gradient changes, the confidence value will be effectively suppressed, preventing noise points, weak texture transitions, and other non-target areas from being misjudged as boundaries.

[0035] The texture similarity for, , For the natural constant An exponential function with base 0. In pixels Pixels within the set neighborhood of the center grayscale value, In pixels The average grayscale value of all pixels within a neighborhood of the center is set. For the pixel point gradient magnitude, This refers to the maximum gradient magnitude of the pixels within the defined neighborhood.

[0036] This design considers both neighborhood grayscale distribution and gradient magnitude information in texture similarity calculation. The deviation between grayscale values ​​and the neighborhood mean is weighted according to the importance of local gradients, and an exponential function is used to perform a non-linear mapping of the weighted differences. This results in regions with more uniform grayscale distribution and smaller gradient changes receiving higher similarity values, while regions with larger grayscale fluctuations and significant gradient changes have significantly lower similarity values. This design not only highlights smooth regions with highly consistent texture structures and effectively reduces the impact of noise points or random intensity fluctuations, but also improves sensitivity to texture continuity while preserving significant edge and detail features.

[0037] The degree of mutation for, , In pixels Target pixels within the set neighborhood of the center gradient direction, For pixels gradient direction, The set neighborhood range is the total number of target pixels, where each target pixel is a pixel whose gradient magnitude is greater than a preset threshold. The set neighborhood range can be 3. 3. Of course, you can also set it according to the actual situation.

[0038] By introducing a neighborhood gradient direction consistency metric, the directional difference between a target pixel and its neighboring pixels with large gradient magnitudes is transformed into a quantifiable abrupt change index, thus more accurately describing the strength of local structural changes. When the gradient direction of a pixel differs little from that of its surrounding target pixels, the abrupt change value is close to the minimum, indicating a smoother structural distribution and better texture continuity within the region. Conversely, when the difference increases significantly, the abrupt change value rises, reflecting the potential presence of obvious edges, sharp corners, cracks, or other abruptly changing structural features. This approach not only effectively suppresses interference from smooth-textured and noisy regions against complex backgrounds but also highlights truly diagnostically significant areas of change in the detection results.

[0039] S3: Obtain the denoised linear guide scan image.

[0040] In a preferred embodiment, an improved anisotropic diffusion filtering algorithm is introduced to denoise the scanned image of the linear guide rail, generating a high-quality denoised image. On one hand, this effectively reduces random noise and stripe artifacts introduced by sensor noise, illumination changes, and high-speed motion acquisition, resulting in a more uniform image background and a significantly improved overall signal-to-noise ratio. On the other hand, it preserves and enhances high-frequency feature information crucial for subsequent detection, such as minor defects, scratches, and wear marks on the linear guide rail surface, making edge contours clearer and texture structures more coherent. Furthermore, this improved diffusion mechanism adaptively adjusts the smoothing range and direction during iteration, effectively avoiding the cumulative effect of noise diffusion to adjacent areas and ensuring the stability and consistency of the denoising results.

[0041] The present invention constructs an artifact boundary confidence score based on texture similarity, abrupt change, and gradient magnitude, and combines it with an improved anisotropic diffusion function to achieve adaptive identification and suppression of artifact regions such as reflections and oil films in linear guide rail scanning images. At the same time, it maintains the original detail protection capability in non-artifact regions. Thus, it can effectively remove artifacts and restore the hidden real texture while preserving the fine structure and texture features of the guide rail surface. Combined with grayscale processing and a real-time computing system, it achieves efficient, accurate, and real-time noise reduction processing of high-resolution scanned images, significantly improving image clarity, detail preservation, and the accuracy and robustness of subsequent defect detection.

[0042] An embodiment of a real-time noise reduction system for scanning images using a linear guide rail:

[0043] like Figure 2 As shown, a block diagram of a real-time denoising system for scanning images using a linear guide rail according to an embodiment of the present invention includes a processor and a memory.

[0044] This invention also provides a real-time noise reduction system for scanning images using a linear guide rail. For example... Figure 2 As shown, the system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a real-time denoising method for linear guideway scanning images according to the present invention.

[0045] The real-time denoising system for scanning images using linear guide rails also includes other components well-known to those skilled in the art, such as communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

[0046] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented by computer-readable / executable instructions stored or otherwise maintained on such a computer-readable medium.

[0047] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise explicitly specified.

[0048] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A real-time denoising method for scanning images of linear guide rails, characterized in that, include: Acquire the scanned image of the linear guide rail to be processed; An improved anisotropic diffusion filtering algorithm is used to denoise the linear guide rail scanning image to be processed, so as to obtain the denoised linear guide rail scanning image. Among them, the improved anisotropic diffusion filtering algorithm includes a diffusion function, which is positively correlated with the initial diffusion function and the artifact boundary confidence of each pixel. The artifact boundary confidence characterizes the confidence that each pixel belongs to an artifact. diffusion function for, , For pixels The confidence level of the artifact boundary. Let be the initial diffusion function; Artifact Boundary Confidence for, , For pixels Texture similarity, For pixels The degree of mutation, For pixels gradient magnitude, , These represent the minimum and maximum gradient magnitudes of pixels in the scanned image of the linear guide rail to be processed. This is the normalization function; Texture similarity for, , For the natural constant An exponential function with base 0. In pixels Pixels within the set neighborhood of the center grayscale value, In pixels The average grayscale value of all pixels within a neighborhood of the center is set. For pixels gradient magnitude, To set the maximum gradient magnitude of pixels within the neighborhood; degree of mutation for, , In pixels Target pixels within the set neighborhood of the center gradient direction, For pixels gradient direction, To set the total number of target pixels within the neighborhood, target pixels are those with a gradient magnitude greater than a preset threshold; the neighborhood range is set to 3.

3.

2. The real-time noise reduction method for scanning images of linear guideways according to claim 1, characterized in that, Use a CCD or CMOS camera to acquire scanned images of the linear guide rail to be processed.

3. The real-time noise reduction method for scanning images of linear guideways according to claim 1, characterized in that, It also includes performing grayscale processing on the scanned image of the linear guide rail to be processed.

4. The real-time denoising method for scanning images of linear guideways according to claim 1, characterized in that, The normalization function is a max-min normalization.

5. A real-time noise reduction system for scanning images using a linear guide rail, characterized in that, It includes a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a real-time denoising method for scanning linear guide rail images as described in any one of claims 1 to 4 is implemented.

Images