Defect Detection System For Prepreg Based On Machine Vision

In order to ensure the performance of the forming components and detect the quality of the prepreg effectively, this paper has designed the prepreg defect detection system based on the machine vision technology, carried on the related research on automatic focusing, image acquisition, preprocessing of prepreg image and prepreg defect recognition technology. The main works of this paper are as follows:First of all, based on the work processing of the machine vision detection, the platform of the prepreg defect detection system has been built. According to the study of image quality in different light intensity, the optimum light intensity standard has been set. In order to obtain a clear image of prepreg, an automatic focusing platform to realize automatic focusing function under a small depth of field has been set up.Secondly, aiming at the problem that the lens which has small depth of field is difficult to focused, this paper has studied the automatic focusing technology of the prepreg image. The quantitative evaluation index of focusing functions has been set and an appropriate focusing function based on this index has been chosen. The mathematical relationship between the focus function value and the coordinate position of the camera has been established by function fitting. According to this function, the positioning method has been adopted to realize the rapid focusing and finally a clear image of prepreg has been got.Also considering the problem that the images captured by the CCD camera have motion blur and low signal-noise ratio(SNR), the image preprocessing module has been divided into image restoration function and noise reduction function. By comparing the restoration effect of inverse filtering method and wiener filtering method, the wiener filtering method has adopted to recovery the motion blurred image of prepreg. According to different type of noise, a variety of filtering methods have been used to reduce the noise and improve the SNR of images, provided good quality of the prepreg image for edge extraction.The prepreg defect can be mainly described by the edge line, so this paper has further studied the prepreg edge location effect of the Canny edge detection algorithm that combined with the automatic threshold. In order to detect the prepreg edge line, the prepreg edge detection algorithm based on the line fitting has been adopted. The interference from the false edge points on the measuring accuracy has been eliminated by citing the straight line fitting method with weigh coefficient. Finally the prepreg edge has been transformed from the image space to the parameter space successfully. Through experiments this algorithm has been verified that it has the same accuracy with the Hough transform but the operation speed is increased by ten times. According to the straight line detected by this method, three characteristic parameters have been set to describe the prepreg defect quantitatively.Finally, through the computer programming language, the prepreg defect visual detection system has been designed. The functions including of automatic focusing, acquisition, processing, analysis and recognition of prepreg image have been realized.