Research On Defects Detecting In Grey Cloth Based On Image Processing

With the development of digital image processing, image processing system have been widely used in many fields. However, how to combine the existent image processing algorithms and construct a system with them effectively is still a problem. In this system, the author carried out the detecting of defects in the grey cloth based with the existent image processing algorithms.Aimed at the characteristics of the images, the author of this paper have been extracted the features of defects in grey cloth effectively after analyzing the images of grey cloth. The fellow steps been applied in the processing with the aid of improving algorithms, then the system get the effects it needed.First, image enhancement methods been applied. After analyzing the advantage and disadvantage of the traditional methods include local averaging, median filter, ideal lowpass filter, Butterworth filter, exponential filter, etc. The system gets the results via applying the image enhancement methods above and compared them. We also explain the principle and implement methods about the novel methods such as morphological filter, wavelet transform, etc. Then the selected algorithms in image enhancement been improved and get the image with good effects and short time consuming.And then, image segmentation methods have been applied to the images in order to get the defects with good effect. After compared and experimented the traditional and novel methods of segmentation such as wavelet transform, we selected one algorithm which can get good effects and short time consuming. After that, we wipe the blobs and connect the defects in the segmented image. We then can get defects image which have good effects via applying morphological algorithms. In the system, we pay more attention to the edge of defects, so we applying Laplacian algorithms to the defects image and get the contour of the defects with good effects. Consider the idiographic features of the image, some algorithms have been optimized and fixed on the parameters in it. Then the good segment effects with short time consuming have been made. In the processing of extracting contour, the system can get the chain code of the contour simultaneous. Then the system can get the shape features parameters (including perimeter, area, and rectangle degree, rotundity degree) of the defects via dealing with chain code. And these parameters can be used as the base of defects recognition in the upper of the system.In the end, the author summarize the present situation of the system, and look into the future about the develop foreground of the system.