Insulation soft rope defect detection system based on machine vision

Abstract

The invention discloses an insulating soft rope defect detection system based on machine vision, relates to the technical field of soft rope detection, solves the problem that different types of defects cannot be accurately distinguished in an original detection mode, and in the aspect of defect classification and tracing, a feature verification end locks an abnormal area through comparison of the abnormal area and a standard image, so that the defect classification and tracing accuracy is improved. The abnormal signal output end is further combined with defect features pre-stored in the cloud database, abnormal features are accurately matched through overlap ratio analysis of gradient difference intervals, the defect types can be determined, storage and tracing of detection data can be achieved through the cloud database, follow-up quality analysis and process optimization are facilitated, and the detection efficiency is improved. And a reliable basis is provided for production improvement and safety application of the insulating soft rope.

Description

Technical Field

[0001] This invention relates to the field of soft rope inspection technology, specifically to a machine vision-based defect detection system for insulating soft ropes. Background Technology

[0002] In fields such as power, industrial lifting, and high-altitude operations, insulated ropes are critical components for ensuring operational safety. Their surface and structural integrity directly affect equipment safety and personnel safety. For example, defects such as breakage, aging, or localized deformation in insulated ropes used in power maintenance can lead to decreased insulation performance and increase the risk of electric shock. Similarly, hidden wear on insulated safety ropes used in lifting operations can cause breakage under load, resulting in serious accidents. Therefore, defect detection in insulated ropes is a core requirement of quality control.

[0003] Traditional defect detection of insulating ropes relies heavily on manual visual inspection, which has significant limitations: Firstly, manual inspection is affected by subjective experience, fatigue, and concentration, resulting in low accuracy in identifying minute defects (such as millimeter-level scratches and local texture anomalies), leading to a high rate of missed and false detections. Secondly, manual inspection is slow (typically only 1-2 meters of rope can be inspected per minute), making it difficult to meet the efficiency requirements of mass production or large-scale operation and maintenance scenarios. Furthermore, it is difficult to standardize inspection standards, which can lead to inconsistent quality judgments due to differences in personnel operation. In addition, manual inspection cannot systematically record and trace defect data, which is not conducive to subsequent quality analysis and process optimization.

[0004] With the development of machine vision technology, automated inspection methods are gradually being applied in the field of industrial quality inspection. However, existing machine vision-based inspection solutions still have shortcomings in the inspection of insulating flexible ropes: some solutions only identify defects through simple edge comparison or grayscale analysis, which is difficult to deal with the irregular contours caused by the flexible shape of the rope; some solutions have limited defect classification capabilities and cannot accurately distinguish different types of defects (such as wear, aging, and breakage), thus limiting the practicality of the inspection results. Therefore, there is an urgent need for an insulating flexible rope defect inspection system that can balance inspection accuracy, efficiency, and defect classification capabilities to solve the pain points of traditional manual inspection and the limitations of existing technologies. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a machine vision-based defect detection system for insulating soft ropes, which solves the problem that the original detection methods cannot accurately distinguish between different types of defects.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a machine vision-based defect detection system for insulating flexible ropes, comprising: The visual monitoring end performs visual monitoring on the insulated soft rope placed on the transmission mechanism and transmits the real-time monitored soft rope image to the image processing end. On the image processing end, the real-time monitored rope image is compared with a preset standard image for feature verification. The contour features present in the image are confirmed, and then the contour features are comprehensively verified to confirm the feature data series associated with the corresponding image. The confirmed feature data series are compared to identify whether the monitored rope image is an abnormal image. The specific method is as follows: The Sobel algorithm was used to confirm the vertical gradient and vertical gradient associated with different pixels within the soft rope image, and the following methods were employed: Confirm the comprehensive gradient associated with the corresponding pixel. Pixels that satisfy the condition that the comprehensive gradient is greater than Y1 are recorded as gradient pixels, where Y1 is a preset value. Otherwise, no calibration is performed. Connect several consecutive gradient pixels to confirm and generate the gradient contour of the corresponding soft rope image. Based on the overall edge of the corresponding soft rope image, the internal center point of the corresponding soft rope image is identified, and the edge point farthest from the internal center point is identified from the overall edge. The edge point is connected to the internal center point to identify several measurement standard lines. Based on the identified gradient contours, the associated regions between adjacent gradient contours are identified and recorded as undetermined regions. The partial line segments of a single measurement standard line located in different undetermined regions are identified, and the line lengths of the partial line segments are identified simultaneously. Based on the identified measurement standard lines, starting from the internal center point and moving towards the edge points, the identified several sets of line lengths are sorted to identify the line length sorting sequence belonging to the corresponding measurement standard line. Then, the length sorting sequence of several measurement standard lines in the current soft rope image is integrated to confirm the feature data column associated with the corresponding image; The same processing method is used to process the preset standard images, and the feature data columns associated with the standard images are recorded synchronously. The standard images are provided by the cloud database. The measurement standard line associated with the soft rope image is compared with the measurement standard line in the standard image. The internal center point is coincident. The line length sorting sequence associated with the coincident measurement standard line is verified. The line length difference at the same sorting position is locked. The different line length differences associated with different line length sorting sequences are averaged to confirm the mean value Jz to be verified. If Jz≤Y2, the monitored soft rope image is marked as an abnormal image; If Jz > Y2, then no marking is performed, and Y2 is the default value.

[0007] The feature verification end compares and verifies the confirmed abnormal image with the standard image to confirm the abnormal regions within the abnormal image. The specific method is as follows: The abnormal image is compared with the standard image. Based on the confirmed internal center point, the abnormal image and the standard image are superimposed. After the superposition is completed, the areas in the abnormal image and the standard image that do not overlap are recorded as abnormal areas, and the confirmed abnormal areas are transmitted to the abnormal signal output terminal. At the abnormal signal output end, based on the confirmed abnormal area, the abnormal area is compared with the defective areas pre-stored in the cloud database to confirm the abnormal characteristics associated with the corresponding abnormal area, and the signal is directly displayed. The specific method is as follows: The defect features in the defective region are identified, and the gradient pixels within the defective region are identified using the Sobel algorithm. The associated comprehensive gradient Z of each gradient pixel is then determined. i , where i represents different gradient pixels. From the confirmed groups of gradient pixels, the gradient difference Cz between adjacent gradient pixels is identified. The different comprehensive gradients associated with adjacent gradient pixels are labeled as TD1 and TD2, and the following formula is used: Cz = |TD1 - TD2|. And from the several sets of gradient differences Cz confirmed in the defect area, the minimum and maximum values ​​are locked to generate a set of difference intervals, and the generated difference intervals are recorded as the defect features of the corresponding defect area. Next, the gradient pixels existing in the abnormal area are confirmed, and the same gradient difference confirmation method is used to lock the gradient interval associated with the corresponding abnormal area. Compare the gradient interval with the defect features of different defect areas to confirm the overlapping range associated with the gradient region and different defect features, and record the percentage of the overlapping range located in the defect features: confirm the interval range of the defect features and the numerical range of the overlapping range, using the formula: numerical range ÷ interval range = percentage value, lock the percentage value, and confirm the different percentage values ​​associated with different defect features. Select the maximum value from the confirmed set of percentage values, record the defect feature associated with the maximum value as an abnormal feature, confirm the abnormal signal associated with the abnormal feature, and directly display the signal.

[0008] Preferably, the visual monitoring terminal is equipped with a shooting time interval T, where T is a preset value.

[0009] This invention provides a machine vision-based defect detection system for insulating flexible ropes. Compared with existing technologies, it has the following advantages: In terms of detection accuracy, the system stably acquires images of the soft rope at preset time intervals through the visual monitoring end. Combined with the Sobel algorithm on the image processing end, it performs fine analysis of the gradient of the pixels, generates gradient contours and constructs feature data columns. Then, by comparing the line length sorting sequence with the standard image and verifying the mean, it can accurately identify abnormal images and effectively capture subtle anomalies on the surface or contour of the insulated soft rope, greatly reducing the risk of missed detection and false detection caused by the limitations of manual visual observation.

[0010] In terms of detection efficiency, the entire process, from image acquisition and processing to anomaly identification, is automated, avoiding the subjectivity and speed bottleneck of manual inspection. The preset shooting time interval (e.g., 2 seconds) can be flexibly adjusted according to actual needs, adapting to batch inspection scenarios and significantly improving the inspection throughput, providing efficient support for the quality control of insulated soft ropes.

[0011] In terms of defect classification and traceability, the feature verification end locks down the abnormal area by comparing the abnormal area with the standard image. The abnormal signal output end further combines the defect features pre-stored in the cloud database and accurately matches the abnormal features through the overlap analysis of the gradient difference interval. This not only clarifies the defect type, but also enables the storage and traceability of detection data by relying on the cloud database, which facilitates subsequent quality analysis and process optimization, and provides a reliable basis for the production improvement and safe application of insulated soft ropes. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the principle framework of the present invention. Detailed Implementation

[0013] 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.

[0014] First Embodiment Please see Figure 1 This application provides a machine vision-based defect detection system for insulating soft ropes, including a visual monitoring end, an image processing end, a feature verification end, a cloud database, and an abnormal signal output end. The visual monitoring end, image processing end, feature verification end, and abnormal signal output end are electrically connected from the output node to the input node in sequence. The cloud database is electrically connected to the input nodes of the feature verification end and the abnormal signal output end, respectively. The visual monitoring end monitors the insulated rope placed on the transmission mechanism and transmits the real-time monitored rope image to the image processing end. The visual monitoring end is set with a shooting time interval T, which is a preset value, usually 2 seconds, and is determined in advance by the operator based on experience. In the image processing section, the real-time monitored rope images are compared with preset standard images to verify their features. The contour features present in the images are confirmed, and then a comprehensive verification of these contour features is performed to confirm the feature data columns associated with the corresponding images. The confirmed feature data columns are then compared to identify whether the monitored rope images are abnormal. The specific identification method is as follows: The Sobel algorithm was used to confirm the vertical gradient and vertical gradient associated with different pixels within the soft rope image, and the following methods were employed: Confirm the comprehensive gradient associated with the corresponding pixel. Pixels that satisfy the condition that the comprehensive gradient is greater than Y1 are recorded as gradient pixels, where Y1 is a preset value and its specific value is determined by the operator based on experience. Otherwise, no calibration is performed. Connect several consecutive gradient pixels to confirm and generate the gradient contour of the corresponding soft rope image. Based on the overall edge of the corresponding soft rope image, the internal center point of the corresponding soft rope image is identified, and the edge point farthest from the internal center point is identified from the overall edge. The edge point is then connected to the internal center point to identify several measurement standard lines (normally, there are generally four sets of corresponding edge points, and the four sets of edge points are connected to the corresponding center point, resulting in four sets of symmetrical measurement standard lines. Based on different measurement standard lines and the corresponding gradient contours, the associated data series can be identified). Based on the identified gradient contours, the associated regions between adjacent gradient contours are identified and recorded as undetermined regions. The line segments of a single measurement standard line located in different undetermined regions are identified, and the line lengths of some line segments are identified simultaneously. Based on the identified measurement standard lines, starting from the internal center point and moving towards the edge points, the identified line lengths are sorted to identify the line length sorting sequence belonging to the corresponding measurement standard line. Then, the length sorting sequence of several measurement standard lines in the current soft rope image is integrated to confirm the feature data column associated with the corresponding image; The same processing method is used to process the preset standard images, and the feature data columns associated with the standard images are recorded synchronously. The standard images are provided by the cloud database. The measurement standard line associated with the soft rope image is compared with the measurement standard line in the standard image. The internal center point is coincident. The line length sorting sequence associated with the coincident measurement standard line is verified. The line length difference at the same sorting position is locked. The different line length differences associated with different line length sorting sequences are averaged to confirm the mean value Jz to be verified. If Jz≤Y2, the monitored soft rope image will be marked as an abnormal image; otherwise, no marking will be made. Y2 is a preset value, and its specific value will be determined by the operator based on experience. Specifically, after each acquired rope image is confirmed, the associated contour points within the image are recorded using a gradient processing algorithm. Based on these recorded contour points, the internal contour lines associated with the rope image are confirmed, revealing gaps between each contour line. The corresponding center point is then identified based on the overall edge region of the rope image. Subsequently, the farthest point is located on the edge based on the center point's position, and the corresponding connecting lines are confirmed based on the identified features. This allows for the combination of multiple sets of measurement lines. The center points are then checked for overlap to identify identical measurement standard lines. The length features associated with these identical standard lines are compared and verified. This comparison and verification process confirms the mean value to be verified, identifies significant differences in the contour features of the corresponding rope images, and promptly identifies abnormal images. These abnormal images are then analyzed and verified.

[0015] The feature verification end compares and verifies the confirmed abnormal image with the standard image, identifies the abnormal regions within the abnormal image, and transmits the identified abnormal regions to the abnormal signal output end. The specific method for identifying abnormal regions is as follows: The abnormal image is compared with the standard image. Based on the confirmed internal center point, the abnormal image and the standard image are superimposed. After the superposition is completed, the areas in the abnormal image and the standard image that do not overlap are recorded as abnormal areas, and the confirmed abnormal areas are transmitted to the abnormal signal output terminal.

[0016] Among them, the abnormal signal output end compares the confirmed abnormal area with the defect area stored in the cloud database to confirm the abnormal features associated with the corresponding abnormal area and directly displays the signal for external personnel to view. Different defect features have preset defect areas, and are associated with different defect features. Therefore, it is necessary to display the signal based on the corresponding comparison process for external relevant personnel to view. The specific method for confirming the abnormal features corresponding to the abnormal region is as follows: The defect features in the defective region are identified, and the gradient pixels within the defective region are identified using the Sobel algorithm. The associated comprehensive gradient Z of each gradient pixel is then determined. i , where i represents different gradient pixels. From the confirmed groups of gradient pixels, the gradient difference Cz between adjacent gradient pixels is identified. The different comprehensive gradients associated with adjacent gradient pixels are labeled as TD1 and TD2, and the following formula is used: Cz = |TD1 - TD2|. And from the several sets of gradient differences Cz confirmed in the defect area, the minimum and maximum values ​​are locked to generate a set of difference intervals, and the generated difference intervals are recorded as the defect features of the corresponding defect area. Next, the gradient pixels existing in the abnormal area are confirmed, and the same gradient difference confirmation method is used to lock the gradient interval associated with the corresponding abnormal area. Compare the gradient interval with the defect features of different defect areas to confirm the overlapping range associated with the gradient region and different defect features, and record the percentage of the overlapping range located in the defect features: confirm the interval range of the defect features and the numerical range of the overlapping range, and use: numerical range ÷ interval range = percentage value to lock the percentage value, and confirm the different percentage values ​​associated with different defect features, select the maximum value from the confirmed set of percentage values, record the defect feature associated with the maximum value as an abnormal feature, confirm the abnormal signal associated with the abnormal feature, and directly display the signal for external personnel to view; Specifically, each defective area has corresponding defective features. In order to better identify the defects in the corresponding rope, the gradient change features associated with the corresponding abnormal area are compared and verified with the confirmed defective features. Based on the comparison and verification process, the overlap of the corresponding abnormal areas is confirmed, thereby locking in the corresponding abnormal signals and displaying them for external personnel to view, so as to facilitate external personnel to take corresponding measures.

[0017] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0018] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A machine vision-based defect detection system for insulating flexible ropes, characterized in that, include: The visual monitoring end performs visual monitoring on the insulated soft rope placed on the transmission mechanism and transmits the real-time monitored soft rope image to the image processing end. On the image processing end, the real-time monitored soft rope image is compared with the preset standard image for feature verification. The contour features existing in the image are confirmed, and then the contour features are comprehensively verified to confirm the feature data column associated with the corresponding image. The confirmed feature data column is compared to identify whether the monitored soft rope image is an abnormal image. The feature verification end compares and verifies the confirmed abnormal image with the standard image to confirm the abnormal area within the abnormal image. At the abnormal signal output end, based on the confirmed abnormal area, the abnormal area is compared with the defective areas pre-stored in the cloud database to confirm the abnormal characteristics associated with the corresponding abnormal area and directly display the signal.

2. The machine vision-based defect detection system for insulating flexible ropes according to claim 1, characterized in that, The visual monitoring terminal is equipped with a shooting time interval T, where T is a preset value.

3. The machine vision-based defect detection system for insulating flexible ropes according to claim 1, characterized in that, The image processing terminal confirms the specific method for identifying the feature data column associated with the corresponding image as follows: The Sobel algorithm was used to confirm the vertical gradient and vertical gradient associated with different pixels within the soft rope image, and the following methods were employed: Confirm the comprehensive gradient associated with the corresponding pixel. Pixels that satisfy the condition that the comprehensive gradient is greater than Y1 are recorded as gradient pixels, where Y1 is a preset value. Otherwise, no calibration is performed. Connect several consecutive gradient pixels to confirm and generate the gradient contour of the corresponding soft rope image. Based on the overall edge of the corresponding soft rope image, the internal center point of the corresponding soft rope image is identified, and the edge point farthest from the internal center point is identified from the overall edge. The edge point is connected to the internal center point to identify several measurement standard lines. Based on the identified gradient contours, the associated regions between adjacent gradient contours are identified and recorded as undetermined regions. The partial line segments of a single measurement standard line located in different undetermined regions are identified, and the line lengths of the partial line segments are identified simultaneously. Based on the identified measurement standard lines, starting from the internal center point and moving towards the edge points, the identified several sets of line lengths are sorted to identify the line length sorting sequence belonging to the corresponding measurement standard line. Then, the length sorting sequence of several measurement standard lines in the current soft rope image is integrated to confirm the feature data column associated with the corresponding image; The preset standard images are processed using the same method, and the feature data columns associated with the standard images are recorded synchronously. The standard images are provided by a cloud database.

4. The machine vision-based defect detection system for insulating flexible ropes according to claim 1, characterized in that, The image processing terminal identifies whether a soft rope image is an abnormal image in the following specific way: The measurement standard line associated with the soft rope image is compared with the measurement standard line in the standard image. The internal center point is coincident. The line length sorting sequence associated with the coincident measurement standard line is verified. The line length difference at the same sorting position is locked. The different line length differences associated with different line length sorting sequences are averaged to confirm the mean value Jz to be verified. If Jz≤Y2, the monitored soft rope image is marked as an abnormal image.

5. The machine vision-based defect detection system for insulating flexible ropes according to claim 4, characterized in that, If Jz > Y2, then no marking is performed, and Y2 is the default value.

6. The machine vision-based defect detection system for insulating flexible ropes according to claim 1, characterized in that, The specific method by which the feature verification terminal confirms abnormal regions is as follows: The abnormal image is compared with the standard image. Based on the confirmed internal center point, the abnormal image and the standard image are superimposed. After the superposition is completed, the areas in the abnormal image and the standard image that do not overlap are recorded as abnormal areas, and the confirmed abnormal areas are transmitted to the abnormal signal output terminal.

7. The machine vision-based defect detection system for insulating flexible ropes according to claim 1, characterized in that, The specific method for displaying the abnormal signal at the output terminal is as follows: The defect features in the defective region are identified, and the gradient pixels within the defective region are identified using the Sobel algorithm. The associated comprehensive gradient Z of each gradient pixel is then determined. i , where i represents different gradient pixels. From the confirmed groups of gradient pixels, the gradient difference Cz between adjacent gradient pixels is identified. The different comprehensive gradients associated with adjacent gradient pixels are labeled as TD1 and TD2, and the following formula is used: Cz = |TD1 - TD2|. And from the several sets of gradient differences Cz confirmed in the defect area, the minimum and maximum values ​​are locked to generate a set of difference intervals, and the generated difference intervals are recorded as the defect features of the corresponding defect area. Next, the gradient pixels existing in the abnormal area are confirmed, and the same gradient difference confirmation method is used to lock the gradient interval associated with the corresponding abnormal area. Compare the gradient interval with the defect features of different defect areas to confirm the overlapping range associated with the gradient region and different defect features, and record the percentage of the overlapping range located in the defect features: confirm the interval range of the defect features and the numerical range of the overlapping range, using the formula: numerical range ÷ interval range = percentage value, lock the percentage value, and confirm the different percentage values ​​associated with different defect features. Select the maximum value from the confirmed set of percentage values, record the defect feature associated with the maximum value as an abnormal feature, confirm the abnormal signal associated with the abnormal feature, and directly display the signal.

Images