Defect Detection Technology Of Fiber Optical Image Transmission Based On Machine Vision

Fiber optical image transmission is a sophisticated optical instrument that appeared in the mid-1950 s.At present,the quality inspection of fiber optical image transmission is mainly done by using microscope or projector.The results of the assessment are subjectively influenced by the staff and the reliability or stability is poor and domestic research on the automatic detection of fiber optical image transmission is less.In the existing research results,several common defects are analyzed in depth and the detection methods of three defects are discussed in detail.Having finished segmentation and quantitative analysis of chickenwire spot and shadow defects.The main work of this paper is as follows:1 According to the characteristics of fiber optical image transmission defects and testing requirements,the hardware system is selected.The high-resolution industrial camera with telephoto industrial lens and LED light to complete the image acquisition of the fiber optical image transmission.2 According to the characteristics of chickenwire and spot defects,a defect segmentation algorithm combining morphological and marker watershed is researched.Firstly,the original image is filtered by morphological reconstruction filter.Then the h-minimum transform is used to eliminate the local minimum of depth less than h and binarize it form the mark image.Finally,the forced minimum transform is used to modify the gradient image so that the local minimum appears only in the mark area and the modified gradient image use watershed transform.The results showed that the method had better segmentation effect on chickenwire and spot defects.3 According to the characteristics of shadow defects,a defect segmentation algorithm based on region growing is researched.The algorithm can automatically generate seed points,avoiding the problem of manually specifying seed points.After the regional growthsegmentation,morphological reconstruction filter is used to eliminate the unclosed voids,So that the defect area to form a complete closed area.The experimental results show that the method can effectively extract the shadow defects.4 The traditional camera calibration method is used to calibrate the relationship between pixel distance and actual distance,and the defects are quantitatively analyzed according to the detection standard of chicken and spot in national standard.