Intelligent identification and evaluation system and method for detecting surface magnetic powder defects

Abstract

The invention discloses an intelligent identification and evaluation system and method for detecting surface magnetic powder defects, and relates to the technical field of image processing. The technical problems that in the prior art, traditional magnetic powder detection depends on manual visual inspection, subjectivity is large, efficiency is low, fatigue is prone to occurring, and the detection result is large in uncertainty and subjectivity due to the fact that the traditional magnetic powder detection is affected by the state, vision and experience of operators are solved. According to the technical scheme provided by the invention, the intelligent identification and evaluation system for detecting the surface magnetic powder defect comprises a module for acquiring a digital image of a detection area; the module is used for preprocessing the digital image; a module for identifying potential defects in the preprocessed digital image; a module for identifying the position of the identified defect; and when the defect reaches a preset alarm condition, an alarm module is triggered. The device is suitable for detection of surface magnetic powder defects.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to detecting surface magnetic powder defects. Background Art

[0002] In the field of industrial non-destructive testing, magnetic particle testing technology is often used to detect defects such as cracks, pores, slag inclusions on the surface and near the surface of ferromagnetic materials. Traditional magnetic particle testing relies on manual visual inspection, which has problems such as high subjectivity, low efficiency, and easy fatigue. Specifically, manual visual inspection requires operators to have high experience and skills, but even so, due to the limitations of the human eye, the test results are still affected by the operator's state, vision and experience, resulting in greater uncertainty and subjectivity in the test results. Operators are prone to fatigue when performing high-intensity testing work for a long time, which further reduces the accuracy and reliability of the test. In addition, the efficiency of manual visual inspection is low, especially when large-scale testing is required, it is difficult to meet the requirements for detection efficiency and speed in industrial production.

[0003] At present, some studies have tried to introduce automation and intelligent technology to improve magnetic particle inspection. For example, machine vision and image processing technology are used to replace manual visual inspection. However, these technologies still have some problems in practical applications. On the one hand, the existing machine vision system has high requirements for image processing algorithms, and complex algorithms often require high-performance hardware support, which increases the cost of the system. On the other hand, the existing system is not accurate in identifying defects, especially in the case of large noise interference, which is prone to false alarms and missed alarms. At the same time, the lack of effective defect identification and classification standards makes it difficult for automated inspection systems to be widely promoted and applied.

[0004] In addition, most of the existing automated testing systems lack real-time and intelligence, and are unable to provide timely feedback and make adjustments during the testing process. The data transmission and storage methods are also not flexible and secure enough, limiting the applicability and reliability of the system in different application scenarios. Therefore, how to further improve the automation level of magnetic particle testing, improve detection accuracy and efficiency, and reduce manual intervention has become a technical problem that needs to be solved in the current field of industrial non-destructive testing. Summary of the invention

[0005] The technical problem that the conventional magnetic particle inspection in the prior art relies on manual visual inspection, which is highly subjective, inefficient, and easy to fatigue, and is affected by the operator's state, vision, and experience, resulting in great uncertainty and subjectivity in the inspection results, is not solved. The technical solution provided by the present invention is:

[0006] An intelligent identification and evaluation method for detecting surface magnetic powder defects, comprising:

[0007] The step of acquiring a digital image of the inspection area;

[0008] A step of preprocessing the digital image;

[0009] The step of identifying potential defects in the pre-processed digital image;

[0010] The step of identifying the location of said identified defect;

[0011] When the defect reaches a preset alarm condition, an alarm is triggered.

[0012] Furthermore, a preferred embodiment is provided, which also includes a step of distinguishing real defects from false defects based on a preset threshold.

[0013] Furthermore, a preferred embodiment is provided, in which the digital image is preprocessed using the OpenCV library, including grayscale processing and Gaussian filtering to remove noise.

[0014] Furthermore, a preferred embodiment is provided, which uses edge detection algorithm and Hough transform to detect straight line features of the pre-processed digital image to identify potential defects.

[0015] Furthermore, a preferred embodiment is provided, which also includes the step of binarizing the digital image.

[0016] Furthermore, a preferred implementation is provided, in which an LSD algorithm is used for crack identification during defect identification.

[0017] Based on the same inventive concept, the present invention also provides an intelligent identification and evaluation system for detecting surface magnetic powder defects, comprising:

[0018] A module for acquiring a digital image of the inspection area;

[0019] A module for preprocessing the digital image;

[0020] A module for identifying potential defects in the pre-processed digital image;

[0021] module for identifying the location of said identified defect;

[0022] When the defect reaches the preset alarm condition, the alarm module is triggered.

[0023] Based on the same inventive concept, the present invention also provides a computer storage medium for storing a computer program. When the computer program is read by a computer, the computer executes the method described.

[0024] Based on the same inventive concept, the present invention also provides a computer, including a processor and a storage medium. When the processor reads a computer program stored in the storage medium, the computer executes the method described.

[0025] Based on the same inventive concept, the present invention also provides a computer program product, which is a computer program. When the computer program is executed, the method described above is implemented.

[0026] Compared with the prior art, the technical solution provided by the present invention is beneficial in that:

[0027] The present invention provides an intelligent recognition and evaluation system for detecting surface magnetic powder defects, which significantly improves the detection efficiency and accuracy by integrating modules such as video acquisition, image processing, algorithm analysis, threshold comparison, identification and alarm. Compared with traditional manual visual inspection, the system reduces subjectivity and human errors through automated processing and is suitable for non-destructive testing of various metal materials.

[0028] The present invention provides an intelligent recognition and evaluation system for detecting surface magnetic powder defects. The video acquisition module uses a camera to capture the digital image of the detection area, thereby achieving full coverage and detail capture of the detection area. Compared with traditional manual recording, the module ensures the integrity and consistency of the data and improves the reliability of detection.

[0029] The present invention provides an intelligent recognition and evaluation system for detecting surface magnetic powder defects. The image processing module uses the OpenCV library to perform image preprocessing, including grayscale processing and Gaussian filtering to remove noise. Compared with other image processing methods, the module's efficient processing capability significantly reduces the interference of noise on the detection results and improves the accuracy of defect recognition.

[0030] The present invention provides an intelligent recognition and evaluation system for detecting surface magnetic powder defects. The algorithm module uses edge detection algorithm and Hough transform to detect straight line features, and uses LSD algorithm to identify potential cracks or defects. Compared with the traditional simple binarization algorithm, this module can extract defect features more accurately, reduce false positives and false negatives, and improve the sensitivity of detection.

[0031] The present invention provides an intelligent recognition and evaluation system for detecting surface magnetic powder defects. The threshold comparison module distinguishes real defects from pseudo defects by adjusting the threshold, line segment length and point distance. Unlike simple algorithms with fixed thresholds, this module can dynamically adjust parameters according to actual conditions, significantly improving the accuracy and reliability of recognition and reducing the occurrence of false alarms.

[0032] The present invention provides an intelligent identification and evaluation system for detecting surface magnetic powder defects. The identification module and the alarm module realize the identification and real-time alarm functions of the identified defect location. Compared with the traditional post-analysis method, the module can timely feedback problems during the detection process, facilitate operators to quickly locate and deal with defects, and improve work efficiency.

[0033] The present invention provides an intelligent recognition and evaluation system for detecting surface magnetic powder defects. The data transmission and storage module realizes remote data transmission and storage of original images and marked images through WiFi technology. Compared with traditional manual recording and local storage, the module provides a more convenient and secure data management method, which is convenient for later data analysis and tracing, and improves the traceability and management level of the detection work.

[0034] The invention provides an intelligent recognition and evaluation system for detecting surface magnetic powder defects, which is suitable for use in the detection of surface magnetic powder defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 The flowchart of the intelligent identification and evaluation method of surface magnetic powder defects;

[0036] Figure 2 This is the logic block diagram of the intelligent identification method for surface magnetic powder defects. DETAILED DESCRIPTION

[0037] In order to make the advantages and benefits of the technical solution provided by the present invention more clearly reflected, the technical solution provided by the present invention is now further described in detail with reference to the accompanying drawings, specifically:

[0038] Embodiment 1: This embodiment provides an intelligent identification and evaluation method for detecting surface magnetic powder defects, including:

[0039] The step of acquiring a digital image of the inspection area;

[0040] A step of preprocessing the digital image;

[0041] The step of identifying potential defects in the pre-processed digital image; the step of marking the location of the identified defects;

[0042] When the defect reaches the preset alarm condition, the alarm is triggered. Specifically:

[0043] Detailed step description:

[0044] Video Collection:

[0045] Use a camera to capture a digital image of the inspection area.

[0046] Output: Image data.

[0047] Image preprocessing:

[0048] Convert color images to grayscale.

[0049] Use Gaussian filtering to remove noise.

[0050] Output: Grayscale processed image.

[0051] Image edge detection:

[0052] Apply an edge detection algorithm (such as Canny algorithm) to the filtered image. Output: edge image.

[0053] Image binarization:

[0054] Convert the extracted edge image into a binary image.

[0055] Output: binary image.

[0056] Line feature detection:

[0057] Use Hough transform (HoughLinesP) to detect straight line features.

[0058] LSD algorithm is used to identify potential cracks or defects.

[0059] Output: Recognized line segment features.

[0060] Defect Identification:

[0061] Image line recognition is performed based on the OpenCV library to draw defect line segments. Output: defect recognition results.

[0062] Defect filtering and identification:

[0063] Adjust the threshold to distinguish real defects from false defects.

[0064] Identify the location of the identified defect.

[0065] Output: Identified and filtered defect data.

[0066] Alarm module:

[0067] Determine whether the alarm condition is met based on the set threshold and trigger the alarm.

[0068] Output: alarm information.

[0069] Data transmission and storage:

[0070] Remote data transmission is achieved through WiFi technology.

[0071] Store original and labeled images.

[0072] Output: stored detection data and images.

[0073] Embodiment 2: This embodiment further limits the intelligent identification and evaluation method for detecting surface magnetic powder defects provided in Embodiment 1, and also includes the step of distinguishing real defects from pseudo defects according to a preset threshold.

[0074] Implementation method three: This implementation method further limits the intelligent identification and evaluation method for detecting surface magnetic powder defects provided in implementation method one, and uses the OpenCV library to preprocess the digital image, including grayscale processing and Gaussian filtering to remove noise.

[0075] Implementation method 4: This implementation method further limits the intelligent identification and evaluation method for detecting surface magnetic powder defects provided in implementation method 1, and uses edge detection algorithm and Hough transform to detect the straight line features of the pre-processed digital image to identify potential defects.

[0076] Embodiment 5: This embodiment further limits the intelligent identification and evaluation method for detecting surface magnetic powder defects provided in Embodiment 1, and also includes the step of binarizing the digital image.

[0077] Embodiment 6: This embodiment further limits the intelligent identification and evaluation method for detecting surface magnetic powder defects provided in embodiment 1. In the process of defect identification, the LSD algorithm is used for crack identification.

[0078] Embodiment 7: This embodiment provides an intelligent identification and evaluation system for detecting surface magnetic powder defects, including:

[0079] A module for acquiring a digital image of the inspection area;

[0080] A module for preprocessing the digital image;

[0081] A module for identifying potential defects in the pre-processed digital image;

[0082] module for identifying the location of said identified defect;

[0083] When the defect reaches the preset alarm condition, the alarm module is triggered.

[0084] Specific implementation method:

[0085] Image acquisition:

[0086] Use the camera to capture digital images of the inspection area and save them to the computer hard drive.

[0087] Grayscale processing:

[0088] Convert color images to grayscale images for easier data analysis.

[0089] Apply a Gaussian filter to remove noise.

[0090] Edge Detection:

[0091] Use the Canny algorithm to obtain image edge information.

[0092] Hough transform (HoughLinesP) is used to detect straight line features.

[0093] Binarization processing:

[0094] The processed images were converted into binary images for further analysis.

[0095] Crack Identification:

[0096] The LSD algorithm in the OpenCV library is used to identify potential cracks or defects.

[0097] Draw the identified defect line segments.

[0098] Defect filtering and identification:

[0099] According to the set threshold, false defects are filtered out and real defects are identified.

[0100] Adjust the threshold, line segment length, and point distance to improve recognition accuracy.

[0101] Real-time alarm:

[0102] Determine whether the identified defects meet the alarm conditions and trigger the alarm.

[0103] Provide real-time alarm information to facilitate operators to respond quickly.

[0104] Data storage and transmission:

[0105] Remote data transmission is achieved through WiFi technology.

[0106] Store original and labeled images for later query and analysis.

[0107] Embodiment 8: This embodiment provides a computer storage medium for storing a computer program. When the computer program is read by a computer, the computer executes the method provided in embodiment 1.

[0108] Embodiment 9: This embodiment provides a computer, including a processor and a storage medium. When the processor reads a computer program stored in the storage medium, the computer executes the method provided in embodiment 1.

[0109] Embodiment 10: This embodiment provides a computer program product, which is a computer program. When the computer program is executed, the method provided in embodiment 1 is implemented.

[0110] Implementation Method XI: Combination Figure 1-2 This embodiment is described in detail by using specific examples to further describe the technical solution provided above. Specifically:

[0111] like Figure 1-2 As shown, in view of the shortcomings of the prior art, the present embodiment aims to provide a surface magnetic powder defect intelligent recognition system, the core of which is to use image processing algorithms to automatically identify and measure the magnetic traces generated in magnetic powder detection, and at the same time have the functions of false defect filtering, automatic crack alarm, wireless data transmission and real-time processing.

[0112] The system comprises a video acquisition module, an image processing module, an algorithm module, a threshold comparison module, an identification module, an alarm module and a data transmission and storage module, and realizes effective recognition and accurate measurement of cracks by optimizing the image processing flow and algorithm parameters.

[0113] Use a camera to capture digital images of the inspection area and save the image data to the computer hard drive.

[0114] The color images are grayscaled, the noise is removed by Gaussian filtering, and the edge detection algorithm is applied to process the filtered images.

[0115] Convert the extracted edge image into a binary image.

[0116] The Canny algorithm is used to obtain image edge information, and the Hough transform (HoughLinesP) is used to detect straight line features and identify potential cracks or defects.

[0117] Crack recognition uses the image line recognition LSD algorithm of OpenCV library 3.0. OpenCV is an open source, cross-platform computer vision and machine learning software library that provides rich image processing and analysis functions.

[0118] The process of crack recognition: image preprocessing, image binarization, image filtering and denoising, extracting feature line segments (straight lines or curves) according to the algorithm, and crack drawing.

[0119] The main process is to grayscale and filter the image first, then generate binary image data, then use corrosion and expansion algorithms to remove noise, use Hough transform to detect edge lines, and then use LSD algorithm to calculate the number of points near the same line segment, use these numbers and proportional operators to match, draw the line segments that meet the requirements, display them on the screen, and use operators to adjust the sensitivity to be detected, which can better solve the problem of crack line segments in dynamic detection and at a faster speed.

[0120] Grayscale The color photos taken at the time are converted into gray images for data analysis. The filtering method uses median filtering, which is a nonlinear filtering. Under certain conditions, median filtering can overcome the blurring of image details caused by linear filters such as minimum mean square filtering and mean filtering, and is most effective in filtering out salt and pepper noise. The filtering principle is to sort the M pixels of the image F(x, y) under the template S (the number of pixels is M), and use the middle value to replace the image pixel value g(x, y) under the center of the template. M is generally an odd number so that there is an intermediate pixel. If M is an even number, the median is the average of the grayscale values ​​of the two intermediate pixels. g(x, y) = Med{F(xi, yj), (i, j∈M)}, where F(xi, yj) represents the grayscale value of the image under the template. The idea of ​​​​Hough transform algorithm:

[0121] Taking line detection as an example, each pixel coordinate point is transformed into a unified metric that contributes to the characteristics of the line. A simple example is as follows: a line in an image is a collection of discrete points. The discrete polar coordinate formula of a line can express the geometric equation of the discrete points of the line.

[0122] Analyze grayscale binary images, filter out noise points, and retain valid data. Adjust the width, length, and threshold of points to improve the accuracy of magnetic trace recognition.

[0123] Regarding the influence of illumination and environment, this implementation adopts the domain value LSD algorithm and Hough line detection algorithm to calculate the dynamic value of the point connection area. The sampling data is not fixed, and there will be jitter display problems.

[0124] The full-screen data is collected using graphics, and then converted into grayscale graphics. The Gaussian blur algorithm is then used to filter out useless noise points, and the grayscale average is taken out. The values ​​above the average are determined as binary sampling points. The LSD algorithm is used to obtain line segments that conform to the lines and mark them in the stored images. The characteristics are that there is no need to set thresholds or edge calculations, and the recognition accuracy will not be improved due to specific parameter adjustments, which reduces the impact of the environment on actual flaw detection. The storage uses a multi-threaded method to synchronously store crack images and original images, and stores video records throughout the process, so that the original images are recorded, the crack images are distinguishable, and the actual video identification is integrated to achieve the purpose of rapid identification alarm prompts and archiving for viewing.

[0125] This method is a method of quickly analyzing tiny data in image recognition, and drawing possible crack line segments to prompt detection, which serves as an auxiliary detection function for magnetic particle flaw detection.

[0126] The individual methods are not new, but their combined application is the advantage of magnetic particle testing's work record keeping and work traceability features, making it well documented.

[0127] From analog manual identification to digital work traces, the difficulty of magnetic particle detection and identification is that the image is affected by the environment and the surface smoothness of the workpiece. The algorithm provides auxiliary display of the features of the connection line, which can facilitate operators to quickly locate and re-inspect.

[0128] Parameters such as the minimum number of curve foci for crack identification and the minimum number of points that make up a straight line are all based on data extraction from simple binary images. This rough solution is only suitable for ordinary demonstrations and is not suitable for actual industrial applications.

[0129] Set the key parameters for crack identification, including the minimum number of curve intersections, the minimum number of points that make up a straight line, and the maximum distance threshold between the two nearest points on a straight line.

[0130] Use the set threshold to determine whether the alarm condition is met, mark the identified crack location, and identify and filter false defects by adjusting the threshold, line segment length and point distance.

[0131] Complete real-time wireless transmission of detection data, store original and marked images, and conduct query and analysis.

[0132] The advantages of this embodiment are:

[0133] Adopt OpenCV library and C++ programming language to realize efficient image processing and pattern recognition, and improve detection speed and accuracy.

[0134] Multi-threaded operation design ensures the real-time performance of image acquisition, processing and alarm feedback.

[0135] The wireless data transmission system enables remote monitoring and storage of test data, facilitating later data analysis and tracing.

[0136] The automatic alarm mechanism combined with threshold adjustment can effectively distinguish real defects from false defects and reduce false alarms and missed alarms.

[0137] Modular design facilitates system upgrade and maintenance.

[0138] Through subsequent deep learning training, the system recognition accuracy will continue to improve and adapt to more complex application scenarios.

[0139] This embodiment solves the problems existing in traditional magnetic particle testing through the above technical solution, provides an automated and intelligent defect identification and evaluation solution, and has broad application prospects.

[0140] The technical solution provided by the present invention is further described in detail above through several specific implementation modes in order to highlight the advantages and benefits of the technical solution provided by the present invention. However, the several specific implementation modes described above are not intended to be used as limitations on the present invention. Any reasonable modification and improvement of the present invention, combination of implementation modes and equivalent substitution within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent identification and evaluation method for detecting surface magnetic powder defects, characterized in that: include: The step of acquiring a digital image of the inspection area; A step of preprocessing the digital image; The step of identifying potential defects in the pre-processed digital image; The step of identifying the location of said identified defect; When the defect reaches a preset alarm condition, an alarm is triggered.

2. The intelligent identification and evaluation method for detecting surface magnetic powder defects according to claim 1 is characterized in that: The method also includes a step of distinguishing real defects from false defects based on a preset threshold.

3. The intelligent identification and evaluation method for detecting surface magnetic powder defects according to claim 1 is characterized in that: The digital image is preprocessed using the OpenCV library, including grayscale processing and Gaussian filtering to remove noise.

4. The intelligent identification and evaluation method for detecting surface magnetic powder defects according to claim 1, characterized in that: Edge detection algorithms and Hough transforms are used to detect straight line features of the pre-processed digital image to identify potential defects.

5. The intelligent identification and evaluation method for detecting surface magnetic powder defects according to claim 1, characterized in that: The method also includes a step of binarizing the digital image.

6. The intelligent identification and evaluation method for detecting surface magnetic powder defects according to claim 1, characterized in that: In the process of defect identification, LSD algorithm is used for crack identification.

7. An intelligent identification and evaluation system for detecting surface magnetic powder defects, characterized in that: include: A module for acquiring a digital image of the inspection area; A module for preprocessing the digital image; A module for identifying potential defects in the pre-processed digital image; module for identifying the location of said identified defect; When the defect reaches the preset alarm condition, the alarm module is triggered.

8. A computer storage medium for storing a computer program, characterized in that: When the computer program is read by a computer, the computer executes the method of claim 1 .

9. A computer, comprising a processor and a storage medium, characterized in that: When the processor reads the computer program stored in the storage medium, the computer executes the method of claim 1 .

10. A computer program product, being a computer program, characterized in that When the computer program is executed, the method of claim 1 is implemented.

Images