Onboard optical fiber end face spot defect detection method, storage medium and equipment

Abstract

The invention discloses a spot defect detection method for an airborne optical fiber end face, which comprises the following steps: firstly, carrying out image acquisition on an optical fiber, then carrying out graying to obtain an image Img1, and then carrying out binaryzation; performing edge detection on the image, extracting the outer edge contour of the optical fiber, taking the extracted outer edge contour as an ROI (Region of Interest), and performing mask operation on a grayscale image in the region to obtain an image Img2; performing automatic threshold binarization processing on the image Img2, and performing binarization segmentation on the image Img2; and automatically extracting the high-brightness light spot in the image Img2 to detect the defect area S'of the high-brightness defect part of the optical fiber, and taking the ratio F of the defect area S 'to the area S of the image Img2 as the defect rate. According to the method, the high-brightness defect part can be accurately extracted, automatic defect detection is realized, and the method has the advantages of high efficiency, simplicity, stability and reliability.

Description

Technical Field

[0001] The present invention belongs to the technical field of optical fiber end face defect detection, and in particular relates to a spot defect detection method, storage medium and equipment for an airborne optical fiber end face. Background Art

[0002] Most of the existing optical fiber measurement and detection are focused on the location, sealing, transmission performance, flow rate, etc. of the optical fiber. The detection methods for the optical fiber end face are also mostly general, mostly a kind of equipment, devices, etc., and the structure and method are relatively complex, prone to errors, resulting in insufficient accuracy and misjudgment, and low automation and efficiency. In the production process of optical fiber end faces, due to the inadequacy of the process flow, it is easy to cause spot defects on the optical fiber end face that are difficult to detect manually. The methods used for the detection of optical fiber end faces are not highly automated and not adaptable enough.

[0003] Existing patent CN106197946B discloses a method, device and system for optical fiber detection. An optical time domain reflectometer is used to detect the waveform generated by the optical fiber link, and the health status of the optical fiber is judged by analyzing and processing the waveform. However, there is no function to detect the end face status. Existing patent CN105652396B discloses a device for optical fiber detection and an optical fiber detection device using the device. The light source is uniformly transmitted through the connecting seat sleeve channel to achieve the function of simultaneously detecting multiple optical fibers. However, it also fails to achieve defect detection of the optical fiber end face.

[0004] Existing patent CN104536090A discloses a fiber end face detection method and a fiber end face polishing and detection device. Its method for detecting damage defects on the fiber end face is ordinary binarization processing. However, ordinary binarization processing cannot realize automated intelligent detection, and it is necessary to set the threshold by oneself, and change the parameters according to different working environments. The use efficiency is not high and the use is difficult. Existing patent CN208672278U discloses a fiber end face detection device and a fiber end face detection grabber. Multiple optical fibers are grasped and detected by a feeder, a grabber, and a discharging device, but it does not mention what method is used to judge whether it is qualified, and the structure is complex and the reliability is not high, which is prone to errors. Summary of the invention

[0005] The purpose of the present invention is to provide a spot defect detection method, storage medium and device for the end face of an airborne optical fiber, aiming to use image processing technology to realize automatic defect detection of the end face of an airborne optical fiber, so as to provide feedback to optimize its manufacturing process and further ensure the quality stability and reliability of the optical fiber.

[0006] The present invention is mainly achieved through the following technical solutions:

[0007] A method for detecting spot defects on an airborne optical fiber end face comprises the following steps:

[0008] Step S1: collecting images of the optical fiber and preprocessing the collected images;

[0009] Step S2: grayscale the preprocessed image to obtain image Img1, and then binarize it;

[0010] Step S3: edge detection is performed on the image to extract the outer edge contour of the optical fiber, and the extracted outer edge contour is used as the ROI region of interest, and the grayscale image in the region is masked to obtain the image Img2;

[0011] Step S4: performing automatic threshold binarization processing on the image Img2 and performing binary segmentation on Img2; automatically extracting the highlight spots in the image Img2 to detect the defect area S' of the highlight defect part of the optical fiber;

[0012] Step S5: The ratio F of the defect area S' to the area S of the image Img2 is taken as the defect rate.

[0013] In order to better implement the present invention, further, in step S4, the automatic threshold binarization processing of the image Img2 includes the following steps:

[0014] Step S41: Since there will be bright abnormal spots in the defective area, set the cluster number K value in the 0-255 grayscale space and the initial normal area cluster center grayscale value C 11 and the gray value C of the abnormal region cluster center 12 ;

[0015] Step S42: Calculate the value of each pixel p in the image Img2 i The distance D from the cluster center k1 ,D k2 :

[0016] D k1 =|P i -C 11 |,

[0017] D k2 =|P i -C 12 |,

[0018] According to the distance, the grayscale values ​​of the pixels Img2 are divided into the normal set Z1 and the abnormal set Z2; 11 Those with are added to the normal set Z1, and those with are added to the abnormal set Z2;

[0019] Step S43: Take the grayscale values ​​of 80% of the pixels closest to the cluster center in the Z1 and Z2 sets respectively, and calculate their average to obtain the new cluster center C. 12 With C 22 ;

[0020] Step S44: Repeat the iteration steps S42 and S43 until the iteration termination condition is met;

[0021] Step S45: Calculate the binarization threshold V:

[0022]

[0023] Where: t is the number of iterations;

[0024] Step S46: Based on the binarization threshold V, perform binary segmentation on the image Img2.

[0025] In order to better implement the present invention, further, in step S41, the cluster number K value is 2, and the initial C 11 =125, C 21 =190.

[0026] In order to better implement the present invention, further, in step S44, the generation termination condition is:

[0027] |C 1t -C 1(t-1) |<1,

[0028] |C 2t -C 2(t-1) |<1.

[0029] In order to better implement the present invention, further, in the step S1, any one or more methods of mean filtering, Gaussian filtering, median filtering, and bilateral filtering are used to perform noise reduction preprocessing on the collected image.

[0030] In order to better implement the present invention, further, in the step S2, the binarization method is any one of the maximum entropy segmentation method, the mean segmentation method, the Otsu method, and the triangular threshold segmentation method.

[0031] In order to better implement the present invention, further, in step S3, the edge detection method is any one of the first-order differential operator, Roberts operator, Prewitt operator, Sobel operator, and Canny operator methods.

[0032] A computer-readable storage medium stores a computer program, which implements the above method when executed by a processor.

[0033] An electronic device comprises a memory and a processor; the memory stores a computer program; the processor is used to execute the computer program in the memory to implement the above method.

[0034] The beneficial effects of the present invention are as follows:

[0035] The present invention uses image processing technology and one-dimensional K-means clustering method to perform automated defect detection, which can effectively avoid the threshold segmentation boundary fuzzy problem caused by general image binarization, so that the highlighted defective part can be extracted more accurately, and has the advantages of high efficiency, simplicity, stability and reliability. The present invention can effectively reduce the difficulty and fatigue strength of manual detection, and achieve automated detection of optical fiber end face defects. Starting from the defect detection of the optical fiber end face, the present invention can effectively optimize and upgrade the production process and ensure product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of the method for detecting spot defects on the end face of an airborne optical fiber in Example 1. DETAILED DESCRIPTION

[0037] Embodiment 1:

[0038] A method for detecting spot defects on an airborne optical fiber end face, such as Figure 1 As shown, the following steps are included:

[0039] Step 1: Install and fix the optical fiber on the clamp, which is fixed on the conveyor belt and driven by the rotating wheel to move.

[0040] Step 2: After completing a single displacement stroke, image acquisition is performed when the optical fiber material reaches the image acquisition station.

[0041] Specifically, the optical fiber to be tested is placed on the clamping end of the clamp, and then the servo motor drives the wheel to rotate, and the wheel drives the conveyor belt to transport the next optical fiber material. Each conveying stroke is a uniform distance between two clamps, ensuring that after the inspection of the previous optical fiber is completed, the next clamp transporting the optical fiber can be moved to the image acquisition station for photo processing and inspection.

[0042] Step 3: Perform noise reduction preprocessing on the collected image. Specifically, the image filtering noise reduction processing generally includes mean filtering, Gaussian filtering, median filtering, bilateral filtering, etc., which are selected according to the characteristics of the detection object in actual conditions.

[0043] Step 4: Grayscale the preprocessed image to obtain image Img, and then binarize it. Specifically, the binarization method can be selected from the maximum entropy segmentation method, mean segmentation method, Otsu method, triangular threshold segmentation method, etc., which is selected according to the characteristics of the detection object in actual conditions.

[0044] Step 5: Perform edge detection on the image to extract edge features of the outer edge of the optical fiber. Specifically, perform edge detection on the image after binarization. The optional methods include first-order differential operator, Roberts operator, Prewitt operator, Sobel operator, Canny operator, etc., which are selected according to the characteristics of the detection object in actual conditions.

[0045] Step 6: Take the extracted outer edge contour as ROI (region of interest), mask the grayscale image Img in the region, and obtain image Img2. Then extract the contour of the outer edge of the optical fiber according to the length feature or circle feature of the outer edge contour of the optical fiber, and then take the area surrounded by the contour as the region of interest, and mask the grayscale image Img in the region to obtain the grayscale image Img2 of the target area.

[0046] Step 7: Since most of the defective points in the optical fiber will have abnormal bright spots, the bright spots in Img2 can be automatically extracted to detect the defective parts of the optical fiber. Therefore, the K-means local clustering binarization method based on regional grayscale value proposed in this invention is used to automatically binarize the image Img2. The details of the automated threshold method are as follows:

[0047] 1) Since there will be bright abnormal spots in the defective area, set the cluster number K value in the 0-255 grayscale space and the initial normal area cluster center grayscale value C 11 Gray value C of the center point of the abnormal regional cluster 12 .

[0048] Preferably, the K value of the cluster center is set to 2, and a binary classification based on the one-dimensional gray value is performed. Due to the highlight feature of the defect point, the normal area cluster center C of the initial cluster center 11 =125, abnormal regional cluster center C 21 =190.

[0049] 2) Calculate the distance D between each pixel in Img2 and the cluster center k1 ,D k2 According to the distance, the grayscale values ​​of the pixels of Img2 are divided into two sets, the normal set Z1 and the abnormal set Z2.

[0050] D k1 =|P i -C 11|,

[0051] D k2 =|P i -C 12 |,

[0052] Compare the two distances and divide the grayscale value of each pixel into two sets according to the closest cluster center. 11 Add Z1 and vice versa.

[0053] 3) Take the grayscale values ​​P of 80% of the pixels closest to the cluster center in Z1 and Z2 respectively 1i , P 2j , find the average value to get the new cluster center C 12 With C 22 .

[0054] Since the grayscale values ​​at the junction of the two sets are very close, there is a fuzzy boundary that is greatly affected by the cluster center. Therefore, when iteratively calculating the cluster center, the grayscale values ​​of 80% of the pixels closest to the cluster center are taken to form new sets Z1′ and Z2′, and then the new cluster center C is calculated. 12 With C 22 .

[0055] P 1i ∈Z1'{P 11 ,P 12 ,…P 1m};

[0056] P 2j ∈Z'2{P 21 ,P 22 ,…P 2n};

[0057]

[0058] 4) Repeat the above steps 2) and 3) and terminate the iteration loop when the cluster centers of adjacent iterations do not change much after a certain number of iterations.

[0059] 5) Calculate the threshold value V obtained by automatic calculation.

[0060] After t iterations, the termination condition is controlled according to actual needs. For example, if the size of the cluster centers calculated between two consecutive times changes very little, the binarization threshold V is calculated as the mean of the final cluster centers:

[0061] C 1t -C 1(t-1) |<1;

[0062] C 2t -C 2(t-1) |<1

[0063]

[0064] Step 8: Use V as the binarization threshold, perform binarization segmentation on Img2, and the ratio F of the area S′ of the highlighted defective part to the area S of image Img2 is the defect rate. Set the threshold of the ratio value according to the process requirements to determine whether the optical fiber is qualified. When the defect rate is greater than the value required for the qualified process, the computer host software outputs an unqualified label and displays a reminder on the display screen, otherwise it indicates that the optical fiber quality is qualified.

[0065] Step 9: When the optical fiber with unqualified test results is transported to the top of the unqualified material box, the clamping end at the end of the clamp is loosened to release the optical fiber and store it in the unqualified material box. Conversely, it will enter the qualified material box in the same way.

[0066] Step 10: When the optical fiber materials in the qualified material box and the unqualified material box are full, a signal is sent to the computer through the attached sensor, and prompts for timely cleaning information are displayed in the computer's host software, and the servo motor is controlled to stop working.

[0067] Specifically, an optical sensor is installed in the optical fiber material box. When the optical fiber material is too much, the feedback signal stops the servo motor operation and prompts the display screen that the material is full and needs to be cleaned up in time. When the cleaning is completed, it can be restarted to resume operation.

[0068] Step 11: After cleaning the material box, restart and continue to load optical fibers for testing until all optical fibers to be tested have been tested.

[0069] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any simple modification or equivalent change made to the above embodiment based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for detecting spot defects on an airborne optical fiber end face, characterized in that: The following steps are involved: Step S1: collecting images of the optical fiber and preprocessing the collected images; Step S2: grayscale the preprocessed image to obtain image Img1, and then binarize it; Step S3: edge detection is performed on the image to extract the outer edge contour of the optical fiber, and the extracted outer edge contour is used as the ROI region of interest, and the grayscale image in the region is masked to obtain the image Img2; Step S4: performing automatic threshold binarization processing on the image Img2 and performing binary segmentation on Img2; automatically extracting the highlight spots in the image Img2 to detect the defect area S' of the highlight defect part of the optical fiber; Step S5: The ratio F of the defect area S' to the area S of the image Img2 is taken as the defect rate.

2. The method for detecting spot defects on an airborne optical fiber end face according to claim 1, characterized in that: In step S4, the automatic threshold binarization processing of the image Img2 includes the following steps: Step S41: Since there will be bright abnormal spots in the defective area, set the cluster number K value in the 0-255 grayscale space and the initial normal area cluster center grayscale value C 11 and the gray value C of the abnormal region cluster center 12 ; Step S42: Calculate the value of each pixel p in the image Img2 i The distance D from the cluster center k1 ,D k2 : D k1 =|P i -C 11 |, D k2 =|P i -C 12 |, According to the distance, the grayscale values ​​of the pixels Img2 are divided into the normal set Z1 and the abnormal set Z2; 11 Those with are added to the normal set Z1, and those with are added to the abnormal set Z2; Step S43: Take the grayscale values ​​of 80% of the pixels in the Z1 and Z2 sets that are closest to the cluster center, and calculate their average to obtain the new cluster center C. 12 With C 22 ; Step S44: Repeat the iteration steps S42 and S43 until the iteration termination condition is met; Step S45: Calculate the binarization threshold V: Where: t is the number of iterations; Step S46: Based on the binarization threshold V, perform binary segmentation on the image Img2.

3. The method for detecting spot defects on an airborne optical fiber end face according to claim 2, characterized in that: In step S41, the cluster number K value is 2, and the initial C 11 =125, C 21 =190.

4. The method for detecting spot defects on an airborne optical fiber end face according to claim 2, characterized in that: In step S44, the generation termination condition is: |C 1t -C 1(t-1) |<1, |C 2t -C 2(t-1) |<1。 5. The method for detecting spot defects on an airborne optical fiber end face according to claim 1, characterized in that: In the step S1, any one or more methods selected from the group consisting of mean filtering, Gaussian filtering, median filtering and bilateral filtering are used to perform noise reduction preprocessing on the collected image.

6. The method for detecting spot defects on an airborne optical fiber end face according to claim 1, characterized in that: In step S2, the binarization method is any one of the maximum entropy segmentation method, the mean segmentation method, the Otsu method, and the triangular threshold segmentation method.

7. The method for detecting spot defects on an airborne optical fiber end face according to claim 1, characterized in that: In step S3, the edge detection method is any one of the first-order differential operator, Roberts operator, Prewitt operator, Sobel operator, and Canny operator methods.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method described in any one of claims 1 to 7 is implemented.

9. An electronic device, characterized in that: The invention comprises a memory and a processor; the memory stores a computer program; the processor is used to execute the computer program in the memory to implement the method according to any one of claims 1 to 7.

Images