Machine Vision-Based Lubrication System Monitoring Method and System

Abstract

This invention relates to the field of image processing, and more specifically, to a machine vision-based method and system for monitoring lubrication systems. The method includes: acquiring an image of abrasive particles in the lubrication system; acquiring the closed edges of the abrasive particle images, defining the area enclosed by these edges as a suspected abrasive particle region; calculating the detail saliency and edge smoothness coefficient of the suspected abrasive particle region; calculating an improved contrast gain based on the detail saliency and edge smoothness coefficient to obtain an improved contrast gain; enhancing the abrasive particle image using the improved contrast gain; identifying the type of abrasive particles using the enhanced abrasive particle image; and determining whether an abnormality has occurred in the lubrication system based on the type of abrasive particles, thereby completing the monitoring of the lubrication system. This invention improves the accuracy of the entire lubrication system monitoring by accurately identifying the type of abrasive particles, ensuring that potential problems in the lubrication system can be detected promptly and accurately.

Description

Technical Field

[0001] This invention relates to the field of image processing. More specifically, this invention relates to a machine vision-based method and system for monitoring lubrication systems. Background Technology

[0002] The lubrication system is the "circulatory system" of mechanical equipment, and its core task is to continuously, appropriately, and cleanly provide lubrication to friction points. By collecting oil samples from the lubrication system and analyzing the types of abrasive particles (such as metal shavings and contaminants) in the oil, the wear condition of the machine or the health of the lubrication system can be determined. The shape, size, and material of the abrasive particles can reflect the type of wear inside the equipment (such as normal wear, fatigue wear, cutting wear, etc.).

[0003] To analyze abrasive particles more accurately, existing technologies perform image processing on the particles (such as noise reduction, contrast enhancement, or edge sharpening) to highlight the detailed features of the abrasive particles (such as surface texture, shape, cracks, etc.) to facilitate subsequent classification or diagnosis.

[0004] However, the morphological characteristics of different abrasive grains vary significantly (e.g., sharp edges of metallic abrasive grains need to be highlighted, while soft textures of fibrous contaminants need to be preserved). Using fixed parameters for image enhancement can lead to two problems: first, some key features of abrasive grains may be over-enhanced (e.g., artifacts appear on metal edges due to excessive sharpening); second, subtle features of other types of abrasive grains may be lost. Furthermore, interference from non-target signals exists during enhancement; irrelevant information such as stray particles in the oil background and image sensor noise may be improperly enhanced, severely interfering with the effective extraction of true abrasive grain features.

[0005] In summary, excessive enhancement or noise will reduce image quality and make it impossible to effectively monitor the lubrication system. Summary of the Invention

[0006] To address the technical problem of the inability to effectively monitor lubrication systems due to the limitations of image enhancement algorithms in lubrication system monitoring, the present invention provides solutions in the following aspects.

[0007] In the first aspect, a machine vision-based lubrication system monitoring method includes:

[0008] Acquire abrasive images of the oil in the lubrication system;

[0009] Obtain the closed edges of the abrasive grain image, and use the area enclosed by them as the suspected abrasive grain region. Calculate the detail saliency and edge smoothness coefficient of the suspected abrasive grain region.

[0010] The improved contrast gain is calculated based on the degree of detail and the edge smoothness coefficient. The improved contrast gain is then used to enhance the abrasive image.

[0011] By using enhanced abrasive images, the types of abrasive particles can be identified, and the type of abrasive particle can be used to determine whether there is an abnormality in the lubrication system, thereby completing the monitoring of the lubrication system.

[0012] This invention first acquires abrasive images of the oil in the lubrication system, enabling a direct observation of the presence of abrasive particles in the oil. Since different types of abrasive particles have different morphologies, sizes, and compositional characteristics, it is necessary to identify the type of abrasive particles to distinguish between normal and abnormal wear. Specifically, by calculating the detail saliency and edge smoothness coefficient of suspected abrasive particle areas, and based on these parameters, calculating an improvement value for contrast gain, the abrasive particle images are enhanced. This process more clearly presents the characteristics of the abrasive particles, reducing misjudgments caused by poor image quality and other factors. Accurate abrasive particle identification enables effective monitoring of the lubrication system.

[0013] Preferably, the enhancement process employs an adaptive contrast enhancement algorithm.

[0014] Preferably, the process of obtaining the salience of the details includes:

[0015] Calculate the texture significance factor and high frequency significance factor of the suspected abrasive region, and use the normalized value of the product of the texture significance factor and the high frequency significance factor as the detail significance of the suspected abrasive region.

[0016] By calculating the texture significance factor (reflecting local texture features) and the high-frequency significance factor (reflecting high-frequency information such as edges and gradients), multi-scale features of the abrasive region can be comprehensively captured. For example, the texture significance factor can highlight texture differences such as scratches and edges on the abrasive surface, while the high-frequency significance factor can enhance the clarity of the abrasive boundary. The product of the two and normalization can effectively distinguish different types of abrasive grains such as cutting abrasive grains and fatigue abrasive grains.

[0017] Preferably, the process of obtaining the edge smoothness coefficient includes:

[0018] Obtain the corner points on the edge of the suspected abrasive grain area, construct its neighborhood with each corner point as the center, obtain the consecutive adjacent edge pixels on both sides of the corner point in the neighborhood to form two edge segments, and calculate the included angle between these two edge segments.

[0019] The included angle is normalized to obtain the roundness contribution of the corner point, and then the mean of the roundness contribution of all corner points is used as the edge smoothness coefficient of the suspected abrasive region.

[0020] Particles produced by normal wear typically have smooth edges, while those produced by abnormal wear (such as fatigue or fracture) may have sharper or irregular edges. By focusing on the corner points (local curvature maxima) on the edges, key geometric features can be captured. The included angle of the corner points directly reflects the local sharpness, and the normalized contribution transforms the discrete corner point characteristics into continuous values, avoiding the excessive influence of a single corner point on the overall evaluation.

[0021] Preferably, the improved contrast gain satisfies the following relationship:

[0022] In the formula, For the improved contrast gain, This is the initial contrast gain. The mean of the degree of detail visibility across all suspected abrasive grain areas. This is the average edge smoothness coefficient of all suspected abrasive regions.

[0023] Preferably, the process of obtaining the texture clarity factor includes:

[0024] Features are extracted from suspected abrasive grain regions, including the mean gradient magnitude and the variance of the gradient direction;

[0025] The product of the mean gradient magnitude and the variance of the gradient direction is used as the texture clarity factor for the suspected abrasive region.

[0026] Gradient magnitude reflects the rate of change in pixel grayscale in an image. In abrasive regions, grayscale changes are usually more pronounced due to factors such as the edges and surface texture of the abrasive grains. Calculating the mean gradient magnitude quantifies the overall intensity of this grayscale change; the gradient direction represents the direction of the grayscale change. Abrasive grain textures often have a certain directionality, such as scratches and texture lines on the abrasive grain surface. Gradient direction variance measures the diversity or consistency of this directionality. If the gradient direction of a region is relatively consistent, it indicates that the region may have a relatively regular texture; if the gradient direction is relatively dispersed, it may indicate that the texture of the region is more complex or that it is not an abrasive grain.

[0027] The texture significance factor is calculated by multiplying the mean gradient magnitude and the variance of the gradient direction, taking into account both the intensity and directionality of the texture. This is because a high mean gradient magnitude alone might simply indicate large local grayscale variations with chaotic directionality, not necessarily representing abrasive grains; conversely, a low gradient direction variance alone, without sufficient gradient magnitude, could simply represent noise or irrelevant texture in the image. By multiplying, the texture significance factor is only significant when both the mean gradient magnitude and the variance of the gradient direction reach a certain level, thus more accurately representing whether the texture features of suspected abrasive grain regions are significant.

[0028] Preferably, the process of obtaining the high-frequency significant factor includes:

[0029] Fourier transform is performed on the suspected abrasive grain region. The low-frequency region at the center of the spectrum is shielded by a mask. The sum of squares of the amplitudes of all pixels in the high-frequency region is calculated as the high-frequency energy. The sum of squares of the amplitudes of the entire spectrum is calculated as the total energy. The ratio of the high-frequency energy to the total energy is used as the high-frequency significance factor of the suspected abrasive grain region.

[0030] Fourier transform converts an image from the spatial domain to the frequency domain. High-frequency components correspond to details such as edges and textures in the image, while low-frequency components correspond to smooth regions (such as the background or large uniform areas). The sum of squares of the amplitudes of all pixels in the high-frequency region is calculated as the high-frequency energy, and the sum of squares of the amplitudes of the entire spectrum is calculated as the total energy. By comparing these, the significance of the high-frequency features of the suspected abrasive region can be quantitatively represented.

[0031] Preferably, the process of obtaining the texture clarity factor includes:

[0032] For each pixel in the suspected abrasive grain area, calculate its LBP value to obtain the LBP histogram. Use the variance of the LBP histogram as the texture significance factor for the suspected abrasive grain area.

[0033] Preferably, the process of obtaining the edge smoothness coefficient includes:

[0034] Edge detection algorithms are used to extract the edges of suspected abrasive regions. The magnitude of the edge gradient is calculated to obtain a gradient image. The variance of the gradient change is calculated based on the gradient image, and the reciprocal of the variance of the gradient change is used as the edge smoothing coefficient.

[0035] Secondly, a machine vision-based lubrication system monitoring system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the machine vision-based lubrication system monitoring method described in any one of the claims is implemented.

[0036] The beneficial effects of this invention are:

[0037] By performing a series of meticulous processing and analysis on abrasive grain images, including obtaining closed edges to identify suspected abrasive grain areas, calculating detail saliency and edge smoothness coefficients, the types of abrasive grains can be identified more accurately. Different types of abrasive grains often correspond to different types of faults or abnormalities in the lubrication system. Accurately identifying the type of abrasive grains can provide a strong basis for judging whether there are abnormalities in the lubrication system, thereby improving the accuracy of judging abnormalities in the lubrication system and effectively avoiding misjudgment or omission. Attached Figure Description

[0038] Figure 1This is a flowchart of steps S1-S4 in the lubrication system monitoring method based on machine vision according to an embodiment of the present invention. Detailed Implementation

[0039] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0040] The application scenario of this invention is: to process the acquired images using improved image enhancement processing technology in order to obtain high-quality images for effective monitoring of the lubrication system.

[0041] Reference Figure 1 The machine vision-based lubrication system monitoring method includes steps S1-S4, as detailed below:

[0042] S1: Obtain an image of the abrasive particles in the lubrication system.

[0043] In one embodiment, the lubricating oil in the lubrication system is sampled and tested periodically. During sampling, the sampling location is kept clean to avoid external contamination. The oil is then filtered through a filter membrane to separate wear particles.

[0044] The acquired wear particles were deposited on a ferrographic slide and then heated to solidify. Ferrographic patterns of the wear particles were obtained using a ferrometer, and images from a ferrographic microscope were acquired using a scanning camera. The acquired images were then processed into grayscale to obtain a wear particle image of the oil in the lubrication system.

[0045] S2: Obtain the closed edges of the abrasive image, and use the area enclosed by them as the suspected abrasive region. Calculate the detail saliency and edge smoothness coefficient of the suspected abrasive region.

[0046] The surface texture of abrasive grains reflects their wear type. Normal wear particles have relatively smooth surfaces with indistinct textures; while abnormal wear particles (such as cutting wear and corrosion wear) have more pronounced surface textures and significant variations in grayscale values. By analyzing texture information, the type of abrasive grain can be determined, thus providing a reference for image enhancement.

[0047] In one embodiment, the Canny edge detection algorithm or the Sobel operator is first used to extract all edges in the abrasive image to locate the contour boundary of the object in the image. Then, by judging whether the first and last coordinates of the edges coincide, the coincident edges are regarded as closed edges. The continuous area enclosed by these closed edges is defined as the suspected abrasive region, thereby identifying independent regions with complete contours in the image.

[0048] Furthermore, HOG (Histogram of Oriented Gradients) feature extraction is performed on each suspected abrasive grain region to obtain two key parameters: the mean gradient magnitude (reflecting the overall intensity of pixel grayscale changes within the region, or the edge sharpness) and the gradient direction variance (characterizing the degree of dispersion of texture direction within the region, or the texture complexity).

[0049] Finally, the product of the calculated mean gradient magnitude and the variance of the gradient direction is used as the texture saliency factor for the suspected abrasive grain region. This texture saliency factor, as a composite index, considers both edge strength and texture complexity, and is used to quantify the texture saliency of the region. The greater the texture saliency, the more obvious the texture of the suspected abrasive grain region. Therefore, when enhancing the image, to prevent over-enhancing the texture, the contrast gain parameter should be decreased; conversely, the contrast gain parameter should be increased.

[0050] It should be noted that high-frequency information in the abrasive grain image also needs to be considered, as high-frequency information typically reflects noise and detail in the image. If there is a high amount of high-frequency information in the abrasive grain region, it indicates that the region contains more noise or detail. During image enhancement, if high-frequency information is not controlled, it can easily lead to over-enhancement, making the noise more noticeable and thus affecting the recognition of abrasive grains.

[0051] A Discrete Fourier Transform (DFT) is performed on the suspected abrasive grain region to convert the spatial domain image into a frequency domain spectrum. In the frequency domain, the low-frequency components (central region) represent the overall structure and contour of the image, while the high-frequency components (edge ​​region) correspond to details, edges, and noise.

[0052] Then, the low-frequency region at the center of the spectrum is shielded by a mask, leaving only the high-frequency region (similar to frequency domain high-pass filtering). The sum of squares of the amplitudes of all pixels in the retained high-frequency region is then calculated as the high-frequency energy. The sum of squares of the amplitudes of the entire spectrum (i.e., low frequency plus high frequency) is then calculated as the total energy.

[0053] The ratio of high-frequency energy to total energy is then used as the high-frequency significance factor for the suspected abrasive grain region. The higher the high-frequency significance factor, the more high-frequency noise or detail is present in the suspected abrasive grain region. Therefore, the contrast gain parameter should be reduced to avoid excessive noise enhancement leading to misidentification of abrasive grains. Conversely, the lower the high-frequency significance factor, the smoother the suspected abrasive grain region is, and the less detail is present. Therefore, the contrast gain parameter should be increased to improve the visibility of details and accurately identify abrasive grains.

[0054] The texture significance factor and high-frequency significance factor obtained through the above calculations can quantify the detailed information of the abrasive grain region.

[0055] Furthermore, the normalized value of the product of the texture significance factor and the high frequency significance factor is used as the detail significance of each suspected abrasive grain region.

[0056] In summary, by combining the texture saliency factor with the high-frequency saliency factor, a detail saliency coefficient is obtained, which dynamically adjusts the contrast gain parameter in the image enhancement algorithm. When the detail saliency coefficient is large, it indicates that there are many details in the abrasive grain area, and the contrast gain needs to be reduced to avoid over-enhancement; conversely, the contrast gain can be appropriately increased to enhance the image contrast.

[0057] In another embodiment, the LBP value is calculated for each pixel in the suspected abrasive region to obtain an LBP histogram, and the variance of the LBP histogram is used as the texture significance factor of the suspected abrasive region.

[0058] The edge characteristics of abrasive grains are an important basis for determining their wear type. The edges of normal wear grains are usually relatively smooth, while the edges of abnormal wear grains (such as cutting wear and corrosion wear) are more irregular and contain more high-frequency information. Therefore, by analyzing the smoothness of the edges, the type of abrasive grain can be identified more accurately.

[0059] In one embodiment, corner points on the edge of the suspected abrasive grain region are obtained using a corner detection algorithm. An n×n neighborhood is constructed centered on each corner point (for example, n is 3; in other embodiments, this can be adjusted according to the actual resolution). Within this neighborhood, consecutive adjacent edge pixels on both sides of the corner point are identified, forming two edge segments. The included angle between these two edge segments is calculated. The more corner points and the smaller the included angle, the sharper the edge and the more detailed information it contains.

[0060] Then, the calculated included angle is normalized to obtain the round contribution of the corner point, which satisfies the following relationship:

[0061]

[0062] In the formula, The contribution of the corner point to the roundness. The angle is the angle between the edge line segments on both sides of the corner point as the center point.

[0063] Next, all corner points in the suspected abrasive grain area were calculated. The values ​​are then averaged to obtain the edge smoothing coefficient for the suspected abrasive grain region. A larger edge smoothing coefficient indicates that the abrasive grain edges are relatively smooth with less detail. In this case, the contrast gain parameter can be appropriately increased to enhance the overall contrast of the image and make the abrasive grain features more obvious. Conversely, a smaller edge smoothing coefficient indicates that the abrasive grain edges contain more detail and the edge variations are more pronounced. In this case, the contrast gain parameter needs to be appropriately decreased to prevent excessive enhancement of details, which could increase noise and affect the abrasive grain recognition effect.

[0064] In another implementation, edge detection algorithms (such as Sobel, Canny, etc.) are first used to extract the edges of suspected abrasive grain regions. The magnitude of the edge gradient is calculated to obtain a gradient image. Statistical analysis is then performed on the gradient image to calculate the variance or standard deviation of the gradient change. The reciprocal of the variance or standard deviation of the gradient change is used as the edge smoothness coefficient. The smaller the variance or standard deviation, the gentler the gradient change and the smoother the edge.

[0065] While the detail sharpness coefficient reflects the texture and high-frequency information of the abrasive grain region, it primarily focuses on the details within the abrasive grains. The edge smoothness coefficient, on the other hand, provides supplementary information from another perspective: the characteristics of the abrasive grain edges. By comprehensively considering both coefficients, the detail information of the abrasive grains can be evaluated more comprehensively, thereby allowing for more precise adjustment of image enhancement parameters.

[0066] The above operations can be used to calculate the detail salience and edge smoothness coefficient of all suspected abrasive areas.

[0067] S3: Calculate the improved value of contrast gain based on the degree of detail and the edge smoothness coefficient to obtain the improved contrast gain, and use the improved contrast gain to enhance the abrasive image.

[0068] Based on the values ​​of the detail clarity coefficient and the edge smoothness coefficient, the contrast gain parameter in the ACE adaptive contrast enhancement algorithm is adjusted to avoid over-enhancement and noise issues when enhancing images containing abrasive particles.

[0069] Specifically, the improved contrast gain satisfies the following relationship:

[0070]

[0071] In the formula, For the improved contrast gain, This is the initial contrast gain. The mean of the degree of detail visibility across all suspected abrasive grain areas. This is the average edge smoothness coefficient of all suspected abrasive regions.

[0072] By improving the contrast gain parameter, targeted enhancement of images containing abrasive particles can be achieved. During the enhancement process, the detailed information and edge features of the abrasive particle surface are fully considered, avoiding the problems of over-enhancement or under-enhancement caused by fixed parameters in traditional image enhancement algorithms. The improved image enhancement algorithm can better highlight the characteristics of abrasive particles, providing clearer and more accurate image evidence for subsequent abrasive particle type identification and lubrication system anomaly determination.

[0073] Then, the abrasive image is enhanced according to the improved contrast gain parameters to obtain the enhanced abrasive image.

[0074] It should be noted that, in this embodiment of the invention, the image enhancement technique used is the ACE adaptive contrast enhancement algorithm.

[0075] S4: Using the enhanced abrasive image, identify the type of abrasive particles, and determine whether there is any abnormality in the lubrication system based on the type of abrasive particles, thereby completing the monitoring of the lubrication system.

[0076] In one embodiment, the enhanced abrasive image obtained in S3 is used to identify the type of abrasive particles and determine whether there is an abnormality in the lubrication system based on the type of abrasive particles, thereby achieving effective monitoring of the lubrication system.

[0077] Specifically, abrasive grain type identification methods (such as feature matching, machine learning, etc.) are used to analyze the enhanced abrasive grain image to identify the types of abrasive grains, including:

[0078] Normal wear particles: such as abrasive particles that are regularly shaped, have smooth surfaces, are small in size, and are evenly distributed, usually indicate that the lubrication system is in normal operating condition.

[0079] Abnormal wear particles: such as irregularly shaped, rough-surfaced, large-sized and unevenly distributed particles, may indicate an abnormality in the lubrication system.

[0080] Common abnormal wear particles include:

[0081] Cutting wear particles: usually in the form of flakes or blocks, with obvious cutting marks on the surface, which may indicate metal contact or insufficient lubrication in the lubrication system.

[0082] Corrosive wear particles: These are usually spherical or irregular in shape, with corrosion marks on the surface, which may indicate the presence of moisture or corrosive substances in the lubrication system.

[0083] Abrasive wear particles: These are usually granular with wear marks on the surface, and may indicate the presence of hard particles in the lubrication system, leading to component wear.

[0084] Fatigue wear particles: usually in the form of flakes or blocks, with fatigue cracks on the surface, which may indicate cyclic loads or insufficient lubrication in the lubrication system.

[0085] Furthermore, the condition of the lubrication system can be determined based on the type and quantity of abrasive particles.

[0086] Specifically, an abnormal threshold for the number of abrasive particles is set. When there are a small number of abnormal abrasive particles but their number does not exceed the set threshold, an early warning signal can be issued to remind staff to pay attention to the operating status of the lubrication system. When the number of abnormal abrasive particles exceeds the set threshold, it is determined that there is an abnormality in the lubrication system, and further inspection and maintenance are required. The oil level and oil quality of the lubrication system can be checked. If the oil level is low, lubricating oil should be added in time, and the lubrication system should be maintained to ensure the normal operation of the motor.

[0087] The system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement the machine vision-based lubrication system monitoring method according to the first aspect of the present invention.

[0088] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0089] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A machine vision-based lubrication system monitoring method, characterized in that, include: Acquire abrasive images of the oil in the lubrication system; Obtain the closed edges of the abrasive grain image, and use the area enclosed by them as the suspected abrasive grain region. Calculate the detail saliency and edge smoothness coefficient of the suspected abrasive grain region. The improved contrast gain is calculated based on the saliency of detail and the edge smoothness coefficient. The improved contrast gain is then used to enhance the abrasive image. By using enhanced abrasive images, the types of abrasive particles can be identified, and the type of abrasive particle can be used to determine whether there is an abnormality in the lubrication system, thereby completing the monitoring of the lubrication system. The process of obtaining the significance of the details includes: Calculate the texture significance factor and high frequency significance factor of the suspected abrasive region, and use the normalized value of the product of the texture significance factor and the high frequency significance factor as the detail significance of the suspected abrasive region.

2. The machine vision-based lubrication system monitoring method according to claim 1, characterized in that, The enhancement process employs an adaptive contrast enhancement algorithm.

3. The lubrication system monitoring method based on machine vision according to claim 1, characterized in that, The process of obtaining the edge smoothness coefficient includes: Obtain the corner points on the edge of the suspected abrasive grain area, construct its neighborhood with each corner point as the center, obtain the consecutive adjacent edge pixels on both sides of the corner point in the neighborhood to form two edge segments, and calculate the included angle between these two edge segments. The included angle is normalized to obtain the roundness contribution of the corner point, and then the mean of the roundness contribution of all corner points is used as the edge smoothness coefficient of the suspected abrasive region.

4. The machine vision-based lubrication system monitoring method according to claim 3, characterized in that, The improved contrast gain satisfies the following relationship: In the formula, For the improved contrast gain, This is the initial contrast gain. The mean of the saliency of detail across all suspected abrasive grain areas. This is the average edge smoothness coefficient of all suspected abrasive regions.

5. The machine vision-based lubrication system monitoring method according to claim 4, characterized in that, The process of obtaining the texture clarity factor includes: Features are extracted from suspected abrasive grain regions, including the mean gradient magnitude and the variance of the gradient direction; The product of the mean gradient magnitude and the variance of the gradient direction is used as the texture clarity factor for the suspected abrasive region.

6. The machine vision-based lubrication system monitoring method according to claim 5, characterized in that, The process of obtaining the high-frequency significant factor includes: Fourier transform is performed on the suspected abrasive grain region. The low-frequency region at the center of the spectrum is shielded by a mask. The sum of squares of the amplitudes of all pixels in the high-frequency region is calculated as the high-frequency energy. The sum of squares of the amplitudes of the entire spectrum is calculated as the total energy. The ratio of the high-frequency energy to the total energy is used as the high-frequency significance factor of the suspected abrasive grain region.

7. The lubrication system monitoring method based on machine vision according to claim 1, characterized in that, The process of obtaining the texture clarity factor includes: For each pixel in the suspected abrasive grain area, calculate its LBP value to obtain the LBP histogram. Use the variance of the LBP histogram as the texture significance factor for the suspected abrasive grain area.

8. The machine vision-based lubrication system monitoring method according to claim 1, characterized in that, The process of obtaining the edge smoothness coefficient includes: Edge detection algorithms are used to extract the edges of suspected abrasive regions. The magnitude of the edge gradient is calculated to obtain a gradient image. The variance of the gradient change is calculated based on the gradient image, and the reciprocal of the variance of the gradient change is used as the edge smoothing coefficient.

9. A machine vision-based lubrication system monitoring system, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the machine vision-based lubrication system monitoring method according to any one of claims 1-8.

Images