An industrial scissor edge precision detection method based on image detection

Abstract

The application relates to the technical field of blade precision detection, and particularly discloses an industrial scissor blade edge precision detection method based on image detection, which comprises the following steps: step S1: placing a to-be-detected scissor blade edge in the center of an acquisition area, collecting blade edge images through a continuously-decreasing exposure time sequence, and obtaining a plurality of image sequences; step S2: determining whether the gray value of each frame and each pixel exceeds an overexposure threshold value, distinguishing between overexposed pixels and non-overexposed pixels, and capturing the critical exposure time stamp when each overexposed pixel is first converted into a non-overexposed state; and step S3: assigning weights to each pixel according to the critical exposure time stamp, constructing a fusion weight map, weighting and fusing the image sequences, and generating a blade edge image without imaging defects; and measuring the blade edge thickness based on the blade edge image, comparing the standard threshold value, and completing the precision qualification determination of the scissor blade edge.

Description

Technical Field

[0001] This invention relates to the field of blade precision detection technology, specifically to an image detection-based method for detecting the precision of industrial scissor blades. Background Technology

[0002] As a core tool in machining and material cutting, the precision of industrial shears directly determines cutting quality and work efficiency, making it a crucial quality control element in the industrial manufacturing sector. During the production and use of shears, precise testing of indicators such as blade flatness, sharpness, and geometric dimensions is necessary to ensure product qualification rates and safe operation.

[0003] However, due to the material properties of metal cutting edges, the surface is highly reflective. When using optical imaging technology for inspection, the strong reflection can easily cause overexposure of the image captured by the image sensor, making the key features and details of the cutting edge obscured by the strong light and unable to be clearly presented. At the same time, the optical distortion caused by high reflectivity will cause the edge imaging to be distorted, making it difficult to accurately extract the geometric parameters of the edge contour, which seriously reduces the detection accuracy and reliability. Summary of the Invention

[0004] The purpose of this invention is to provide an image detection-based method for detecting the precision of industrial scissor blades, thereby solving the following technical problems.

[0005] The objective of this invention can be achieved through the following technical solutions: An image detection-based method for detecting the precision of industrial scissor blades includes the following steps: Step S1: Place the blade of the scissors to be detected at the center of the image acquisition area, and illuminate the surface of the blade of the scissors to be detected with a preset constant light source; set a continuously decreasing exposure time sequence, and acquire images of the surface of the blade of the scissors to be detected with the exposure time sequence to obtain an image sequence; Step S2: For any frame image in the image sequence, determine whether the gray value of each pixel exceeds the preset overexposure gray value threshold. If it exceeds the overexposure gray value threshold, the pixel is recorded as an overexposure pixel; otherwise, it is recorded as a non-overexposure pixel. Obtain the critical exposure timestamp corresponding to the first time the overexposure pixel becomes a non-overexposure pixel in the image sequence. Step S3: Assign weight values ​​to each pixel according to the critical exposure timestamp, and construct a fusion weight map with the same size as the image; perform weighted fusion on the image sequence according to the fusion weight map to obtain the cutting edge image; obtain the cutting edge thickness value according to the cutting edge image, and determine whether the cutting edge of the scissors to be detected is qualified based on the cutting edge thickness value.

[0006] As a further aspect of the present invention: the process of setting a continuously decreasing exposure time series includes: During the formal testing of the scissor blade, a sample of the scissor blade is selected, and a sample image of the scissor blade sample is acquired with a preset initial exposure time. The number of all overexposed and non-overexposed images in the sample image is obtained, and the ratio between overexposed and non-overexposed images is obtained. If the ratio is greater than a preset first ratio threshold, the initial exposure time is gradually reduced according to a preset attenuation factor until the ratio is less than or equal to the first ratio threshold, and the gradually reduced initial exposure time is recorded as the minimum effective exposure time. If the ratio is greater than a preset second ratio threshold, the initial exposure time is gradually increased according to a preset gain factor until the ratio is less than or equal to the second ratio threshold, and the gradually increased initial exposure time is recorded as the maximum effective exposure time. Between the minimum effective exposure time and the maximum effective exposure time, several intermediate exposure times are generated by arithmetic progression. The total number of intermediate exposure times, the minimum effective exposure time, and the maximum effective exposure time constitute a continuously decreasing exposure time sequence.

[0007] As a further aspect of the present invention, the process of setting the continuously decreasing exposure time series further includes: All intermediate exposure times, the minimum effective exposure time, and the maximum effective exposure time within the exposure time series are recorded as exposure times. Each exposure time within the exposure time series is sequentially numbered, and the time interval between any two adjacent exposure times is recorded.

[0008] As a further aspect of the present invention: the process of obtaining the critical exposure timestamp includes: When acquiring the image sequence, the timestamp corresponding to the first exposure time in the exposure time sequence is recorded, and the exposure time corresponding to the image where the overexposed pixel first becomes a non-overexposed pixel in the image sequence is obtained, which is denoted as the critical exposure time; based on the critical exposure time and the time interval, the timestamp corresponding to the critical exposure time is obtained, which is denoted as the critical exposure timestamp.

[0009] As a further aspect of the present invention: the process of assigning weight values ​​to each pixel based on the critical exposure timestamp includes, for any pixel, recording the critical exposure timestamp of the pixel as t, and recording the total number of images in the image sequence as M, then the weight value of the pixel W = (Mt) / M.

[0010] As a further aspect of the present invention: the process of constructing a fusion weight map with the same size as the image includes: Create a blank image with the same size and number of pixels as any image in the image sequence, and set the pixel values ​​of all pixels in the blank image to 0; traverse all pixels in the blank image, obtain the weight values ​​of the pixels at the corresponding positions in the image, set the weight values ​​as the pixel values ​​of the pixels in the blank image, and eliminate the abrupt changes in the pixel values ​​of all pixels in the blank image based on a two-dimensional Gaussian smoothing filter method to obtain a fused weight map.

[0011] As a further aspect of the present invention: the process of weighted fusion of the image sequence to obtain the blade edge image includes: In the image sequence, the image corresponding to the maximum effective exposure time is denoted as the longest exposure image I. long The image corresponding to the minimum effective exposure time is denoted as the shortest exposure image I. short For any pixel K in the image, obtain the pixel value W in the fusion weight map, and obtain the grayscale value I of the pixel in the longest exposure image. long (K), and obtain the grayscale value I of the pixel in the shortest exposure image. short If (K), then the true grayscale value of the pixel is obtained as R(K) = W × I. long (K) + (1-W) × I short (K); then, in the final fusion weight map, the blade image is formed according to the true grayscale value of each pixel in the fusion weight map.

[0012] As a further aspect of the present invention: the process of obtaining the blade thickness value based on the blade image includes performing sub-pixel level edge extraction on the blade image to obtain two contour edge lines of the blade to be detected, and obtaining the average distance between the two contour edge lines, which is recorded as the blade thickness value.

[0013] The beneficial effects of this invention are: Unlike traditional single-exposure methods that struggle to capture details in both bright and dark areas, this invention utilizes the longest exposure image to obtain complete texture and contour information in low-reflectivity areas, while simultaneously using the shortest exposure image to preserve detailed features in high-reflectivity areas. Then, by analyzing the critical exposure time index pixel-by-pixel and constructing a fusion weight map, the two types of images are adaptively weighted and fused to ultimately generate a high dynamic range image with no overexposed areas and continuously sharp edges. This process does not rely on expensive polarization optics or complex optical path designs; it can be achieved using only a conventional industrial camera and a controllable light source, significantly reducing hardware costs and system complexity.

[0014] Furthermore, the construction of the fusion weight map in this invention adopts a linear mapping formula negatively correlated with the critical exposure time index, and Gaussian smoothing filtering is performed in the spatial domain. This not only ensures a smooth transition between high-reflectivity and low-reflectivity areas and avoids splicing artifacts, but also effectively suppresses noise introduced by pixel-level independent judgment. This invention can control the positioning error of the cutting edge to the sub-pixel level, and the repeatability of cutting edge thickness measurement is better than traditional methods. It realizes the quantitative judgment of the sharpness and integrity of the scissors. The judgment logic is clear and can be automatically executed. It is suitable for rapid online detection on industrial production lines, effectively improving detection efficiency and consistency.

[0015] In summary, this invention solves the problem of imaging highly reflective metals while taking into account detection accuracy, system cost, and engineering practicality, and has significant technological advancements. Attached Figure Description

[0016] The invention will now be further described with reference to the accompanying drawings.

[0017] Figure 1 This is a schematic diagram of the steps of an image detection-based method for detecting the precision of industrial scissor blades according to the present invention. Figure 2 This is a flowchart of an image detection-based method for detecting the precision of industrial scissor blades according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 As shown, this invention is an image detection-based method for detecting the precision of industrial scissor blades, comprising the following steps: Step S1: Place the blade of the scissors to be detected at the center of the image acquisition area, and illuminate the surface of the blade of the scissors to be detected with a preset constant light source; set a continuously decreasing exposure time sequence, and acquire images of the surface of the blade of the scissors to be detected with the exposure time sequence to obtain an image sequence; In a preferred embodiment of the present invention, the image acquisition of the image acquisition area is based on an image acquisition device. The blade of the scissors to be detected is placed horizontally with its blade facing upward in the image acquisition area, and the extension direction of the blade of the scissors to be detected is parallel to the horizontal pixel row direction of the image acquisition device. The blade of the scissors to be detected is controlled to move along the X-axis and Y-axis directions of the horizontal plane until the tip of the blade of the scissors to be detected is located at the cross mark of the image acquisition area, thus completing the centering. It is worth noting that two sets of strip LED light sources symmetrically arranged on both sides of the blade edge are used as the constant illumination source. The illumination plane of the two sets of strip LED light sources forms an incident angle of 45°±5° with the surface of the blade edge of the scissors to be tested. Furthermore, the two sets of light sources have the same color temperature and equal light intensity, so as to form a uniform diffuse reflection illumination field on the surface of the blade edge of the scissors to be tested, eliminate the directional highlights caused by a single-direction light source, and suppress unidirectional specular reflection to the greatest extent, so that the blade edge surface can obtain uniform diffuse reflection illumination. In a preferred embodiment of the present invention, the process of setting the continuously decreasing exposure time sequence includes: During the formal testing of the scissor blade, a sample of the scissor blade is selected, and a sample image of the scissor blade sample is acquired with a preset initial exposure time. The number of all overexposed and non-overexposed images in the sample image is obtained, and the ratio between overexposed and non-overexposed images is obtained. If the ratio is greater than a preset first ratio threshold, the initial exposure time is gradually reduced according to a preset attenuation factor until the ratio is less than or equal to the first ratio threshold, and the gradually reduced initial exposure time is recorded as the minimum effective exposure time. If the ratio is greater than a preset second ratio threshold, the initial exposure time is gradually increased according to a preset gain factor until the ratio is less than or equal to the second ratio threshold, and the gradually increased initial exposure time is recorded as the maximum effective exposure time. Between the minimum effective exposure time and the maximum effective exposure time, several intermediate exposure times are generated by arithmetic progression, and the total number of intermediate exposure times, the minimum effective exposure time, and the maximum effective exposure time constitute a continuously decreasing exposure time sequence. The process of setting a continuously decreasing exposure time series also includes: All intermediate exposure times, the minimum effective exposure time, and the maximum effective exposure time within the exposure time series are recorded as exposure times. Each exposure time within the exposure time series is sequentially numbered, and the time interval between each two adjacent exposure times is recorded. It is understood that the image corresponding to the maximum effective exposure time in the image sequence contains complete details of the low reflectivity area of ​​the blade edge, but the high reflectivity area is overexposed; the image corresponding to the minimum effective exposure time in the image sequence has clear details in the high reflectivity area, but the low reflectivity area is too dark. Step S2: For any frame image in the image sequence, determine whether the gray value of each pixel exceeds the preset overexposure gray value threshold. If it exceeds the overexposure gray value threshold, the pixel is recorded as an overexposure pixel; otherwise, it is recorded as a non-overexposure pixel. Obtain the critical exposure timestamp corresponding to the first time the overexposure pixel becomes a non-overexposure pixel in the image sequence. In a preferred embodiment of the present invention, the process of determining whether an overexposed pixel becomes a non-overexposed pixel for the first time in the image sequence includes: Based on the number of each exposure time in the exposure time sequence, an index 1, 2, ..., M is set for each image in the image sequence, where M is the total number of images, and index 1 corresponds to the longest exposure image and index M corresponds to the shortest exposure image. For any pixel K in the image, starting from the gray value in the corresponding index 1 image, the gray values ​​at the same pixel position in the images of index 2, index 3 up to index M are traversed sequentially. During the traversal, if the gray value corresponding to the current index K is greater than the overexposure threshold, the traversal continues to the next index. If the gray value corresponding to the current index K is less than or equal to the overexposure threshold for the first time, the traversal stops, and the index K is recorded as the critical exposure time index of pixel K. If, after traversing all images from index 1 to M, the grayscale value of the pixel in all images is greater than the overexposure threshold, then its critical exposure time index is recorded as M+1, indicating that the pixel is still an overexposed pixel even under the shortest exposure; if the grayscale value of the pixel in all images is less than or equal to the overexposure threshold, then its critical exposure time index is recorded as 0, indicating that the pixel has never been overexposed. In a preferred embodiment of the present invention, the process of obtaining the critical exposure timestamp includes: When acquiring the image sequence, the timestamp corresponding to the first exposure time in the exposure time sequence is recorded, and the exposure time corresponding to the image where the overexposed pixel first becomes a non-overexposed pixel in the image sequence is obtained, which is denoted as the critical exposure time; based on the critical exposure time and the time interval, the timestamp corresponding to the critical exposure time is obtained, which is denoted as the critical exposure timestamp. Step S3: Assign weight values ​​to each pixel according to the critical exposure timestamp, and construct a fusion weight map with the same size as the image; perform weighted fusion on the image sequence according to the fusion weight map to obtain the blade edge image; obtain the blade edge thickness value according to the blade edge image, and determine whether the blade edge of the scissors to be detected is qualified based on the blade edge thickness value; In a preferred embodiment of the present invention, the process of assigning a weight value to each pixel according to the critical exposure timestamp includes: for any pixel, the critical exposure timestamp of the pixel is recorded as t, and the total number of images in the image sequence is recorded as M, then the weight value of the pixel is W=(Mt) / M. It can be understood that when t=0, W=1, indicating that the pixel has never been overexposed and the image corresponding to the maximum effective exposure time is used completely; when t=M, W=0, indicating that the pixel is between the maximum and minimum effective exposure time; when t=M+1, W=0, indicating that the pixel is still overexposed even at the minimum effective exposure time and cannot recover the effective grayscale value, so it is assigned a weight of 0 and triggers supplemental lighting or re-acquiring the image sequence; Furthermore, there is a negative correlation between the weight value and the critical exposure timestamp. The smaller the critical exposure timestamp, the closer the weight value is to 1, and the main contribution during fusion comes from the long exposure image; the larger the index, the smaller the weight, and the main contribution during fusion comes from the short exposure image. In a preferred embodiment of the present invention, the process of constructing a fusion weight map with the same size as the image includes: Create a blank image with the same size and number of pixels as any image in the image sequence, and set the pixel values ​​of all pixels in the blank image to 0; traverse all pixels in the blank image, obtain the weight values ​​of the pixels at the corresponding positions in the image, set the weight values ​​as the pixel values ​​of the pixels in the blank image, and eliminate the abrupt changes in the pixel values ​​of all pixels in the blank image based on a two-dimensional Gaussian smoothing filter method to obtain a fused weight map; It is understandable that the determination of the critical exposure timestamp of each pixel is independent. Adjacent pixels may have different critical exposure timestamps due to small grayscale fluctuations. Therefore, a two-dimensional Gaussian smoothing filter method is used to eliminate abrupt changes. In a preferred embodiment of the present invention, the process of weighted fusion of the image sequence to obtain the blade edge image includes: In the image sequence, the image corresponding to the maximum effective exposure time is denoted as the longest exposure image I. long The image corresponding to the minimum effective exposure time is denoted as the shortest exposure image I. short For any pixel K in the image, obtain the pixel value W in the fusion weight map, and obtain the grayscale value I of the pixel in the longest exposure image. long (K), and obtain the grayscale value I of the pixel in the shortest exposure image. short If (K), then the true grayscale value of the pixel is obtained as R(K) = W × I. long (K) + (1-W) × I short(K); then finally, in the fusion weight map, the blade image is formed according to the real gray value of each pixel in the fusion weight map; It is understandable that when W=1, R(K)=I long R(K) represents the pixel using the grayscale value of the longest exposure image; when W=0, R(K)=I short (K) indicates that the pixel uses the grayscale value of the image with the shortest exposure. In a preferred embodiment of the present invention, the process of obtaining the blade thickness value based on the blade image includes: performing subpixel-level edge extraction on the blade image to obtain two contour edge lines of the blade to be detected, and obtaining the average distance between the two contour edge lines, which is recorded as the blade thickness value. It is understandable that the details of the high reflectivity area in the obtained blade thickness value come from the shortest exposure image, and the details of the low reflectivity area come from the longest exposure image. Furthermore, the different areas are naturally stitched together through the continuous transition of weight values, without any stitching marks. In a preferred embodiment of the present invention, the process of determining whether the blade edge of the scissors to be tested is qualified based on the blade edge thickness value includes: The blade thickness value is compared with a preset standard thickness range threshold. If the blade thickness value belongs to the standard thickness range threshold, the precision of the blade of the scissors to be tested is qualified; otherwise, the precision of the blade of the scissors to be tested is unqualified. The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A method for detecting the precision of industrial scissor blades based on image detection, characterized in that, Includes the following steps: Step S1: Place the blade of the scissors to be detected at the center of the image acquisition area, and illuminate the surface of the blade of the scissors to be detected with a preset constant light source; set a continuously decreasing exposure time sequence, and acquire images of the surface of the blade of the scissors to be detected with the exposure time sequence to obtain an image sequence; Step S2: For any frame image in the image sequence, determine whether the gray value of each pixel exceeds the preset overexposure gray value threshold. If it exceeds the overexposure gray value threshold, the pixel is recorded as an overexposure pixel; otherwise, it is recorded as a non-overexposure pixel. Obtain the critical exposure timestamp corresponding to the first time the overexposure pixel becomes a non-overexposure pixel in the image sequence. Step S3: Assign weight values ​​to each pixel according to the critical exposure timestamp, and construct a fusion weight map with the same size as the image; perform weighted fusion on the image sequence according to the fusion weight map to obtain the cutting edge image; obtain the cutting edge thickness value according to the cutting edge image, and determine whether the cutting edge of the scissors to be detected is qualified based on the cutting edge thickness value.

2. The method for detecting the precision of industrial scissor blades based on image detection according to claim 1, characterized in that, In step S1, the process of setting the continuously decreasing exposure time series includes: During the formal testing of the scissor blade, a sample of the scissor blade is selected, and a sample image of the scissor blade sample is acquired with a preset initial exposure time. The number of all overexposed and non-overexposed images in the sample image is obtained, and the ratio between overexposed and non-overexposed images is obtained. If the ratio is greater than a preset first ratio threshold, the initial exposure time is gradually reduced according to a preset attenuation factor until the ratio is less than or equal to the first ratio threshold, and the gradually reduced initial exposure time is recorded as the minimum effective exposure time. If the ratio is greater than a preset second ratio threshold, the initial exposure time is gradually increased according to a preset gain factor until the ratio is less than or equal to the second ratio threshold, and the gradually increased initial exposure time is recorded as the maximum effective exposure time. Between the minimum effective exposure time and the maximum effective exposure time, several intermediate exposure times are generated by arithmetic progression. The total number of intermediate exposure times, the minimum effective exposure time, and the maximum effective exposure time constitute a continuously decreasing exposure time sequence.

3. The method for detecting the precision of industrial scissor blades based on image detection according to claim 2, characterized in that, In step S1, the process of setting the continuously decreasing exposure time series also includes: All intermediate exposure times, the minimum effective exposure time, and the maximum effective exposure time within the exposure time series are recorded as exposure times. Each exposure time within the exposure time series is sequentially numbered, and the time interval between any two adjacent exposure times is recorded.

4. The method for detecting the precision of industrial scissor blades based on image detection according to claim 3, characterized in that, In step S2, the process of obtaining the critical exposure timestamp includes: When acquiring the image sequence, the timestamp corresponding to the first exposure time in the exposure time sequence is recorded, and the exposure time corresponding to the image where the overexposed pixel first becomes a non-overexposed pixel in the image sequence is obtained, which is denoted as the critical exposure time; based on the critical exposure time and the time interval, the timestamp corresponding to the critical exposure time is obtained, which is denoted as the critical exposure timestamp.

5. The method for detecting the precision of industrial scissor blades based on image detection according to claim 1, characterized in that, In step S3, the process of assigning weight values ​​to each pixel according to the critical exposure timestamp includes: for any pixel, the critical exposure timestamp of the pixel is recorded as t, and the total number of images in the image sequence is recorded as M, then the weight value of the pixel is W=(Mt) / M.

6. The method for detecting the precision of industrial scissor blades based on image detection according to claim 1, characterized in that, In step S3, the process of constructing a fusion weight map with the same size as the image includes: Create a blank image with the same size and number of pixels as any image in the image sequence, and set the pixel values ​​of all pixels in the blank image to 0; traverse all pixels in the blank image, obtain the weight values ​​of the pixels at the corresponding positions in the image, set the weight values ​​as the pixel values ​​of the pixels in the blank image, and eliminate the abrupt changes in the pixel values ​​of all pixels in the blank image based on a two-dimensional Gaussian smoothing filter method to obtain a fused weight map.

7. The method for detecting the precision of industrial scissor blades based on image detection according to claim 2, characterized in that, In step S3, the process of weighted fusion of the image sequence to obtain the cutting edge image includes: In the image sequence, the image corresponding to the maximum effective exposure time is denoted as the longest exposure image I. long The image corresponding to the minimum effective exposure time is denoted as the shortest exposure image I. short For any pixel K in the image, obtain the pixel value W in the fusion weight map, and obtain the grayscale value I of the pixel in the longest exposure image. long (K), and obtain the grayscale value I of the pixel in the shortest exposure image. short If (K), then the true grayscale value of the pixel is obtained as R(K) = W × I. long (K) + (1-W) × I short (K); then, in the final fusion weight map, the blade image is formed according to the true grayscale value of each pixel in the fusion weight map.

8. The method for detecting the precision of industrial scissor blades based on image detection according to claim 1, characterized in that, In step S3, the process of obtaining the blade thickness value based on the blade image includes performing subpixel-level edge extraction on the blade image to obtain two contour edge lines of the blade to be detected, and obtaining the average distance between the two contour edge lines, which is recorded as the blade thickness value.

Images