Design And Implementation Of A Capsule Defect Detection System Based On Computer Vision

In the competitive market environment, industrial testing is becoming more and more important, also the requirements for industrial production testing are also much higher, increasing cost of inputs promotes the detection of our country enters into rapid development period, industrial inspection mode and Efficiency has also been significantly improved.In the period of industrial detection, defect detection occupy a major part and defect detection is an important application for image processing and computer vision. Defect detection usually refers to the detection for the surface defects of goods, the workpiece surface spots, pits, scratches, color, defects and other defects will be detected using the advanced machine vision technology of surface detection. Currently, many domestic and foreign software companies have developed classes of detection software, the system can automatically detect the defects according to the technical settings, and identify defective parts, it also could automatically be sorted and removed according to your need, instead of doing artificial recognition, that system will improve production efficiency.The design of this thesis for the capsule defect detection system regardless of hardware or software has been improved over existing systems. On detection machine, streamlined disc-shaped capsule slot detection has been redesigned, detection camera will have four, two of them will take the front of five capsules, the other two will take the top single shot capsule, there will be 12 images are waiting for detection of each capsule. On the software side, the pretreatment process designs of the capsule and detection algorithms are redesigned, ensuring the image only includes the capsule when detecting. The detection algorithm and image processing techniques such as extraction, segmentation, edge detection, dilation, erosion, binarization will be combined to process the image of the capsule, then determine whether the capsule has the defect or not according to the outcome of the above. If the capsule has the defect, mark the categories and locate the placement of the defect, then the defect is marked visually in the saved image, eventually control the machine to remove the defect. On detection algorithm side, the methods to detect the black spots, holes, bubbles are optimized, operation is simplified. By the way respectively counting the color of cap, sleeve, body combined area of three parts for the capsule. Based on the color mixed batch of capsules defect detection; increase translucent capsule defect detection.Compared with other capsule defect detection system, it has four characteristics. First, the hardware configuration of the machine-to-machine has been improved, increasing the number of cameras to ensure that the full range of capsules to be taken to detect. Second, the detection algorithm innovation, optimization, can be more accurate detection of various defects. Third, the detection speed of the machine is to reach 80,000 per hour, regardless of the accuracy and speed can meet the testing requirements. Fourth, the system has translucent capsule for detection capabilities.