Image-based rope quality detection method and detection system

Abstract

The invention provides an image-based rope quality detection method and system, and the method comprises the steps: shooting a rope in a straightening state, and obtaining a video or image of the rope; extracting a to-be-detected rope image from the video or image of the rope, and then converting the extracted rope image into a grey-scale map and performing histogram equalization; and identifying a rope contour and a rope strand contour in the rope image after histogram equalization, and jointly judging the quality of the rope based on the information of the rope contour and the rope strand contour. According to the rope quality detection method and system based on the image, a deep learning method does not need to be used for training, the rope quality can be detected only through a traditional visual method, data labeling is not needed, the generalization ability is high, and the rope quality detection method and system can be applied to various scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of rope quality detection, in particular to a rope quality detection method and system based on images. BACKGROUND

[0002] In many industrial and logistics fields, ropes and strands are key load-bearing elements, widely used in cargo hoisting, binding and traction, etc. However, due to long-term use and external environmental influence, the strands may have quality problems such as wear, breakage or loosening, which directly affects the safety and service life of the rope. The traditional rope quality detection method mainly relies on manual visual inspection, which is time-consuming and labor-intensive, and is easily affected by subjective judgment and visual fatigue, resulting in inaccurate and unstable detection results.

[0003] With the rapid development of computer vision and image processing technology, image-based rope quality detection methods have emerged. This method captures the image of the rope using a high-resolution camera, and then uses computer algorithms to process and analyze the image, thereby realizing automatic detection of rope quality. However, the existing image-based rope quality detection methods have the following problems: 1. Image quality is greatly affected by lighting conditions. Under different lighting conditions, the color and texture features of the rope may change, resulting in unstable detection results; 2. The rope may have multiple types of defects, and different defects have different manifestations, which requires the detection algorithm to have strong generalization ability, increasing the complexity of model training; 3. High-quality training data set is crucial for deep learning methods, but in rope quality detection, the acquisition of labeled data may be very difficult and time-consuming.

[0004] Therefore, it is an urgent problem to propose a new image-based rope quality detection method. SUMMARY

[0005] In view of this, the present application provides an image-based rope quality detection method and system to solve the problems existing in the prior art.

[0006] In one aspect, the present application provides an image-based rope quality detection method, comprising:

[0007] Photographing the rope in a straightened state to obtain a video or image of the rope;

[0008] Extracting the rope image to be detected from the video or image of the rope, and then converting the extracted rope image to a grayscale image and performing histogram equalization;

[0009] The rope profile and the strand profile in the rope image after histogram equalization are recognized, and the quality of the rope is determined based on information of the rope profile and the strand profile;

[0010] The quality of the rope is determined based on information of the rope profile and the strand profile, and the determination specifically includes:

[0011] The determination of whether the track lines of the two edges of the rope profile meet the equidistance line requirement within a preset first deviation range is based on the rope profile, and if the track lines meet the equidistance line requirement, it is determined that the quality of the rope is not problematic, otherwise, it is determined that the quality of the rope is problematic.

[0012] The strand profile deviation feature value is calculated based on the strand profile, and compared with a preset deviation feature threshold value, and if the strand profile deviation feature value is less than the preset deviation feature threshold value, it is determined that the quality of the rope is not problematic, otherwise, it is determined that the quality of the rope is problematic.

[0013] The angle between the strand edge and the rope edge is calculated based on the rope profile and the strand profile, and the deviation of the angle is counted, and if the deviation is less than a preset angle deviation threshold value, it is determined that the quality of the rope is not problematic, otherwise, it is determined that the quality of the rope is problematic.

[0014] Preferably, the rope profile and the strand profile in the rope image after histogram equalization are recognized by using a method combining blob analysis and morphology.

[0015] Further preferably, the determination of whether the track lines of the two edges of the rope profile meet the equidistance line requirement within a preset first deviation range includes the following steps:

[0016] The track lines of the two edges of the rope profile are obtained based on the rope profile.

[0017] It is determined whether the track lines of the two edges are equidistance lines within a preset first deviation range.

[0018] Further preferably, the determination method of the equidistance line is as follows:

[0019] Suppose a planar curve C, whose parametric equation is r(t) = (x(t), y(t)), then the parametric equation of the curve Ck equidistant to the curve C is:

[0020] rk(t) = (x(t) + k*Nx(t), y(t) + k*Ny(t));

[0021] In the formula, k is a distance constant, Nx(t) and Ny(t) are components of the normal vector of the curve at point t.

[0022] Further preferably, the calculation of the strand profile deviation feature value based on the strand profile includes the following steps:

[0023] Based on the strand contour, the complete strand edge trajectory in the image is determined and a closing operation is performed on the trajectory to close the internal region of the trajectory.

[0024] The variances of the area, roundness, contour length, and concavity / convexity of all closed regions are calculated, and the strand profile deviation features are calculated based on the obtained variances. The calculation method for the strand profile deviation features is as follows:

[0025] Features=αD(area)+βD(circularity)+γD(contlength)+δD(convexity);

[0026] In the formula, D(area) represents the area variance, D(circularity) represents the roundness variance, D(contlength) represents the contour length variance, D(convexity) represents the concavity variance, α, β, γ and δ represent the weights of the area variance, roundness variance, contour length variance and concavity variance respectively, and α+β+γ+δ=1.

[0027] Further preferred, calculating the angle between the strand edge and the rope edge based on the rope profile and strand profile, and statistically analyzing the deviation of the angle, includes the following steps:

[0028] The angle between each strand edge and the rope edge is determined based on the rope profile and strand profile.

[0029] The included angles are statistically determined and their variance D(θ) is calculated.

[0030] The present invention also provides an image-based rope quality inspection system, comprising:

[0031] A video or image acquisition unit is used to capture images of a rope in a taut state to obtain video or images of the rope.

[0032] The preprocessing unit is used to extract the rope image to be detected from the video or image of the rope. Then, the extracted rope image is converted into a grayscale image and histogram equalization is performed.

[0033] The quality inspection unit is used to identify the rope outline and strand outline in the rope image after histogram equalization, and jointly determine the quality of the rope based on the information of the rope outline and strand outline.

[0034] The quality inspection unit includes:

[0035] The first detection unit is used to determine whether the trajectory lines of the two edges of the rope profile meet the equidistant line requirements within a preset first deviation range based on the rope profile. If they meet the requirements, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty.

[0036] The second detection unit is used to calculate the deviation feature value of the rope strand profile based on the rope strand profile and compare it with the preset deviation feature threshold. If the deviation feature value is less than the preset deviation feature threshold, it is determined that the rope quality is fine; otherwise, it is determined that the rope quality is faulty.

[0037] The third detection unit is used to calculate the angle between the edge of the rope strand and the edge of the rope based on the rope profile and the strand profile, and to count the deviation of the angle. If the deviation is less than a preset angle deviation threshold, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty.

[0038] Preferably, the first detection unit includes:

[0039] The first acquisition unit is used to obtain the trajectory lines of the two edges of the rope profile based on the rope profile;

[0040] The first determination unit is used to determine whether the trajectory lines of the two edges are equidistant lines within a preset first deviation range.

[0041] More preferably, the second detection unit includes:

[0042] The second processing unit is used to determine the complete edge trajectory of the rope strands in the image based on the rope strand contour and to perform a closing operation on the trajectory to close the internal region of the trajectory.

[0043] The second calculation unit is used to calculate the variance of the area, roundness, contour length, and concavity / convexity of all closed regions, and to calculate the strand profile deviation features based on the obtained variances. The calculation method for the strand profile deviation features is as follows:

[0044] Features=αD(area)+βD(circularity)+γD(contlength)+δD(convexity);

[0045] In the formula, D(area) represents the area variance, D(circularity) represents the roundness variance, D(contlength) represents the contour length variance, D(convexity) represents the concavity variance, α, β, γ and δ represent the weights of the area variance, roundness variance, contour length variance and concavity variance respectively, and α+β+γ+δ=1.

[0046] More preferably, the first detection unit includes:

[0047] The third processing unit is used to determine the angle between the edge of each strand and the edge of the rope based on the rope profile and the strand profile.

[0048] The third calculation unit is used to statistically determine the included angle and calculate the variance D(θ) of the included angle.

[0049] This invention provides an image-based rope quality detection method and system that can detect the quality of ropes in a taut state. First, video or images are acquired and preprocessed (converted to grayscale and histogram equalization is performed). Then, the rope outline and strand outlines are identified, and the rope quality is determined based on the features of these outlines. This method does not require deep learning training; it can detect rope quality using only traditional visual methods. It does not require data annotation, has strong generalization ability, and can be applied to various scenarios. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart of the image-based rope quality detection method provided by the present invention. Detailed Implementation

[0052] The present invention will be further described below with reference to specific embodiments.

[0053] like Figure 1 As shown, this invention provides an image-based rope quality detection method, comprising the following steps:

[0054] S1: Take a picture of the rope when it is taut to obtain a video or image of the rope;

[0055] Preferably, a high-resolution camera is used to image the rope in real time, ensuring that the distance between the camera and the rope is appropriate to ensure that a clear image of the rope is captured;

[0056] S2: Extract the rope image to be detected from the video or image of the rope. Then, convert the extracted rope image into a grayscale image and perform histogram equalization to make the boundaries between the rope, background, and rope strands clear. The formula for histogram equalization is as follows:

[0057] P(rk) = nk / MN;

[0058] Sk=T(rk)=(L-1)*∑ni / MN;

[0059] In the formula, P(rk) represents the probability of gray level rk, nk represents the number of times gray level rk occurs, MN represents the total number of pixels in the image, Sk represents the new gray level mapped to gray level rk, L represents the maximum gray level, and ni represents the number of times gray level i occurs.

[0060] S3: Identify the rope outline and strand outline in the rope image after histogram equalization, and jointly determine the quality of the rope based on the information of the rope outline and strand outline;

[0061] Among them, the method of combining blob analysis and morphology is used to identify the rope contour and strand contour in the rope image after histogram equalization.

[0062] Specifically, determining the quality of a rope based on information about its outline and strand outline includes:

[0063] S31: Based on the rope profile, determine whether the trajectory lines of the two edges of the rope profile meet the equidistant line requirements within the preset first deviation range. If they do, determine that the rope quality is fine; otherwise, determine that the rope quality is faulty.

[0064] S32: Calculate the deviation feature value of the rope strand profile based on the rope strand profile and compare it with the preset deviation feature threshold. If the deviation feature value is less than the preset deviation feature threshold, it is determined that the rope quality is fine; otherwise, it is determined that the rope quality is faulty.

[0065] S33: Calculate the angle between the edge of the rope strand and the edge of the rope based on the rope profile and the strand profile, and count the deviation of the angle. If the deviation is less than the preset angle deviation threshold, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty.

[0066] In step S31, determining whether the trajectory lines of the two edges of the rope profile meet the equidistant line requirement within a preset first deviation range based on the rope profile includes the following steps:

[0067] S311: Obtain the trajectory lines of the two edges of the rope profile based on the rope profile;

[0068] S312: Determine whether the trajectory lines of the two edges are equidistant lines within a preset first deviation range, wherein the method for determining equidistant lines is as follows:

[0069] Suppose there is a plane curve C with the parametric equation r(t) = (x(t), y(t)). Then the parametric equation of a curve Ck that is equidistant from curve C is:

[0070] rk(t)=(x(t)+k*Nx(t), y(t)+k*Ny(t));

[0071] In the formula, k is the distance constant, and Nx(t) and Ny(t) are the components of the normal vector of the curve at point t;

[0072] In step S32, the calculation of the strand profile deviation characteristic value based on the strand profile includes the following steps:

[0073] S321: Determine the complete edge trajectory of the rope strands in the image based on the rope strand contour and perform a closing operation on the trajectory to close the internal region of the trajectory;

[0074] S322: Calculate the variances of the area, roundness, contour length, and concavity / convexity of all closed regions, and calculate the strand profile deviation features based on the calculated variances. The calculation method for the strand profile deviation features is as follows:

[0075] Features=αD(area)+βD(circularity)+γD(contlength)+δD(convexity);

[0076] In the formula, D(area) represents the area variance, D(circularity) represents the roundness variance, D(contlength) represents the contour length variance, and D(convexity) represents the concavity variance. α, β, γ, and δ represent the weights of the area variance, roundness variance, contour length variance, and concavity variance, respectively. α+β+γ+δ=1, preferably α=0.43, β=0.16, γ=0.19, and δ=0.22.

[0077] After obtaining the characteristic value of the strand profile deviation, the rope quality can be determined by comparing the characteristic value of the strand profile deviation with the preset characteristic threshold of the strand profile deviation.

[0078] If this step determines that there is a problem with the quality of the rope, it indicates that the strands may have wear, breakage, or other quality issues.

[0079] In step S33, calculating the angle between the edge of the rope strand and the edge of the rope based on the rope profile and the strand profile, and calculating the deviation of the angle, includes the following steps:

[0080] S331: Determine the angle between the edge of each strand and the edge of the rope based on the rope profile and strand profile;

[0081] S332: Statistically determine the included angle and calculate the variance D(θ) of the included angle;

[0082] Then, the variance D(θ) of the included angle is compared with the preset included angle deviation threshold. If the value of D(θ) is less than the preset included angle deviation threshold, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty.

[0083] The present invention also provides an image-based rope quality inspection system, characterized in that it includes:

[0084] A video or image acquisition unit is used to capture images of a rope in a taut state to obtain video or images of the rope.

[0085] The preprocessing unit is used to extract the rope image to be detected from the video or image of the rope. Then, the extracted rope image is converted into a grayscale image and histogram equalization is performed.

[0086] The quality inspection unit is used to identify the rope outline and strand outline in the rope image after histogram equalization, and jointly determine the quality of the rope based on the information of the rope outline and strand outline.

[0087] The quality inspection unit includes:

[0088] The first detection unit is used to determine whether the trajectory lines of the two edges of the rope profile meet the equidistant line requirements within a preset first deviation range based on the rope profile. If they meet the requirements, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty.

[0089] The second detection unit is used to calculate the deviation feature value of the rope strand profile based on the rope strand profile and compare it with the preset deviation feature threshold. If the deviation feature value is less than the preset deviation feature threshold, it is determined that the rope quality is fine; otherwise, it is determined that the rope quality is faulty.

[0090] The third detection unit is used to calculate the angle between the edge of the rope strand and the edge of the rope based on the rope profile and the strand profile, and to count the deviation of the angle. If the deviation is less than a preset angle deviation threshold, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty.

[0091] The first detection unit includes:

[0092] The first acquisition unit is used to obtain the trajectory lines of the two edges of the rope profile based on the rope profile;

[0093] The first determination unit is used to determine whether the trajectory lines of the two edges are equidistant lines within a preset first deviation range.

[0094] The second detection unit includes:

[0095] The second processing unit is used to determine the complete edge trajectory of the rope strands in the image based on the rope strand contour and to perform a closing operation on the trajectory to close the internal region of the trajectory.

[0096] The second calculation unit is used to calculate the variance of the area, roundness, contour length, and concavity / convexity of all closed regions, and to calculate the strand profile deviation features based on the obtained variances. The calculation method for the strand profile deviation features is as follows:

[0097] Features=αD(area)+βD(circularity)+γD(contlength)+δD(convexity);

[0098] In the formula, D(area) represents the area variance, D(circularity) represents the roundness variance, D(contlength) represents the contour length variance, D(convexity) represents the concavity variance, α, β, γ and δ represent the weights of the area variance, roundness variance, contour length variance and concavity variance respectively, and α+β+γ+δ=1.

[0099] The first detection unit includes:

[0100] The third processing unit is used to determine the angle between the edge of each strand and the edge of the rope based on the rope profile and the strand profile.

[0101] The third calculation unit is used to statistically determine the included angle and calculate the variance D(θ) of the included angle.

[0102] This image-based rope quality inspection method and system can inspect the quality of ropes in a taut state. First, it acquires video or images and preprocesses them (converting them to grayscale and performing histogram equalization). Then, it identifies the rope outline and strand outlines, and determines the rope quality based on the features of these outlines. This method does not require deep learning training; it can detect rope quality using only traditional visual methods. It does not require data annotation, has strong generalization ability, and can be applied to various scenarios.

[0103] It should be noted that the purpose of disclosing the embodiments is to help further understand the present invention; however, those skilled in the art will understand that various substitutions and modifications are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, the present invention should not be limited to the content disclosed in the embodiments, and the scope of protection of the present invention is defined by the claims.

Claims

1. An image-based method for rope quality detection, characterized in that, include: The rope is filmed while it is taut to obtain video or images of the rope; Extract the rope image to be detected from the video or image of the rope. Then, convert the extracted rope image into a grayscale image and perform histogram equalization. The rope profile and strand profile are identified in the rope image after histogram equalization, and the quality of the rope is jointly determined based on the information of the rope profile and strand profile. Specifically, determining the quality of a rope based on information about its outline and strand outline includes: Based on the rope profile, determine whether the trajectory lines of the two edges of the rope profile meet the equidistant line requirements within the preset first deviation range. If they do, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty. The deviation feature value of the rope strand profile is calculated based on the rope strand profile and compared with a preset deviation feature threshold. If the deviation feature value is less than the preset deviation feature threshold, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty. The angle between the edge of the rope strand and the edge of the rope is calculated based on the rope profile and the strand profile, and the deviation of the angle is statistically analyzed. If the deviation is less than a preset angle deviation threshold, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty.

2. The image-based rope quality detection method according to claim 1, characterized in that, The rope contour and strand contour in the rope image after histogram equalization were identified using a combination of blob analysis and morphology.

3. The image-based rope quality detection method according to claim 1, characterized in that, Determining whether the trajectory lines of the two edges of the rope profile meet the equidistant line requirement within a preset first deviation range based on the rope profile includes the following steps: The trajectory lines of the two edges of the rope profile are obtained based on the rope profile; Determine whether the trajectory lines of the two edges are equidistant lines within a preset first deviation range.

4. The image-based rope quality detection method according to claim 3, characterized in that, The method for determining equidistant lines is as follows: Suppose there is a plane curve C with the parametric equation r(t) = (x(t), y(t)). Then the parametric equation of a curve Ck that is equidistant from curve C is: rk(t)=(x(t)+k*Nx(t),y(t)+k*Ny(t)); In the formula, k is the distance constant, and Nx(t) and Ny(t) are the components of the normal vector of the curve at point t.

5. The image-based rope quality detection method according to claim 1, characterized in that, Calculating the characteristic value of the deviation of the strand profile based on the strand profile includes the following steps: Based on the strand contour, the complete strand edge trajectory in the image is determined and a closing operation is performed on the trajectory to close the internal region of the trajectory. The variances of the area, roundness, contour length, and concavity / convexity of all closed regions are calculated, and the strand profile deviation features are calculated based on the obtained variances. The calculation method for the strand profile deviation features is as follows: Features=αD(area)+βD(circularity)+γD(contlength)+δD(convexity); In the formula, D(area) represents the area variance, D(circularity) represents the roundness variance, D(contlength) represents the contour length variance, D(convexity) represents the concavity variance, α, β, γ and δ represent the weights of the area variance, roundness variance, contour length variance and concavity variance respectively, and α+β+γ+δ=1.

6. The image-based rope quality detection method according to claim 1, characterized in that, Calculating the angle between the edge of the strand and the edge of the rope based on the rope profile and strand profile, and calculating the deviation of the angle, includes the following steps: The angle between each strand edge and the rope edge is determined based on the rope profile and strand profile. The included angles are statistically determined and their variance D(θ) is calculated.

7. An image-based rope quality inspection system, characterized in that, include: A video or image acquisition unit is used to capture images of a rope in a taut state to obtain video or images of the rope. The preprocessing unit is used to extract the rope image to be detected from the video or image of the rope. Then, the extracted rope image is converted into a grayscale image and histogram equalization is performed. The quality inspection unit is used to identify the rope outline and strand outline in the rope image after histogram equalization, and jointly determine the quality of the rope based on the information of the rope outline and strand outline. The quality inspection unit includes: The first detection unit is used to determine whether the trajectory lines of the two edges of the rope profile meet the equidistant line requirements within a preset first deviation range based on the rope profile. If they meet the requirements, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty. The second detection unit is used to calculate the deviation feature value of the rope strand profile based on the rope strand profile and compare it with the preset deviation feature threshold. If the deviation feature value is less than the preset deviation feature threshold, it is determined that the rope quality is fine; otherwise, it is determined that the rope quality is faulty. The third detection unit is used to calculate the angle between the edge of the rope strand and the edge of the rope based on the rope profile and the strand profile, and to count the deviation of the angle. If the deviation is less than a preset angle deviation threshold, the rope quality is determined to be fine; otherwise, the rope quality is determined to be faulty.

8. The image-based rope quality inspection system according to claim 7, characterized in that, The first detection unit includes: The first acquisition unit is used to obtain the trajectory lines of the two edges of the rope profile based on the rope profile; The first determination unit is used to determine whether the trajectory lines of the two edges are equidistant lines within a preset first deviation range.

9. The image-based rope quality inspection system according to claim 7, characterized in that, The second detection unit includes: The second processing unit is used to determine the complete edge trajectory of the rope strands in the image based on the rope strand contour and to perform a closing operation on the trajectory to close the internal region of the trajectory. The second calculation unit is used to calculate the variance of the area, roundness, contour length, and concavity / convexity of all closed regions, and to calculate the strand profile deviation features based on the obtained variances. The calculation method for the strand profile deviation features is as follows: Features=αD(area)+βD(circularity)+γD(contlength)+δD(convexity); In the formula, D(area) represents the area variance, D(circularity) represents the roundness variance, D(contlength) represents the contour length variance, D(convexity) represents the concavity variance, α, β, γ and δ represent the weights of the area variance, roundness variance, contour length variance and concavity variance respectively, and α+β+γ+δ=1.

10. The image-based rope quality inspection system according to claim 7, characterized in that, The first detection unit includes: The third processing unit is used to determine the angle between the edge of each strand and the edge of the rope based on the rope profile and the strand profile. The third calculation unit is used to statistically determine the included angle and calculate the variance D(θ) of the included angle.

Images