Study Of Image Processing For A Defect Detection System Of The Inner Surface Of Pipeline

Pipeline transport is not only the important component of comprehensive transportation for a national economy, but also one of the characteristics which can measure this country’s transportation and energy are advanced or not. But due to the influence of such factors as natural conditions, transportation medium and man-made destruction, some damage problems like corrosion and crack are much more likely to happen. So it is very important in checking the spatial distribution and deformation of pipelines and detecting defects by reconstructing inner surface’s three-dimensional information of pipeline.Paper consider large diameter pipeline in the petroleum chemical industry field as tested object. Through studying the related theory about binocular stereo vision system, the ring cross section method based on circle structured light vision detection technology is used. After designing and producing prototype equipment of detection system which contains two CMOS, a simple and effective calibration method is found and the measure precision we get is0.11cm.The focus work of this paper is image processing which consist of analog processing and actual processing. Analog processing tests whether the method we use in this paper is feasible. And on this basis the concrete steps about actual processing of image are divided into image preprocessing, image binaryzation, image thinning, image transition and scaling, image rotation and image fusion. Finally, author finishes three-dimensional shape reconstruction including defect information for the inner surface of pipeline and the maximum error of measurement we obtain is2.20%. The results show that this method has a certain innovation and practicability, it provides a new train of thought about defect detection for the internal wall of a pipeline.