Research On Parameters Measurement Of Aerospace Parts Based On Image Processing Technology

As we all known, aerospace industrial component is the important foundationof space industry development in our country. And it is of great importance. But fora long time, the inspection of some parts parameters rely on artificial to complete,and the problems of inconsistent quality and low inspection efficiency exist.Therefore it is urgently need to improve detection means to solve the traditionalmanual detection problems such as low efficiency and low accuracy.Taking the two parts of geometry measurement accuracy and efficiency intoconsideration, combining with the existing image processing technology, this paperbases on the aerospace parts for high resolution CCD image, and makes furtherresearch about the key technologies of computer vision detection theory.This paper studies the research status of the field of parts parameter detectionat home and abroad, discusses the importance of computer vision technology in theaerospace parts detection, illustrates the feasibility, necessity and practical value ofvision inspection technology, and discusses kinds of algorithms about the detectionof arc and line. The paper discusses the normal processing of digital imagetechnology, analyzes the sources and properties of noise in digital images, givescommon filter algorithms and discusses the technology of image edge detectiondeeply.This paper studies the basic principle of Hough transform and RandomizedHough transform, lists the steps of circular arc and straight line detection. What ismore, it analyzes the advantages and disadvantages of the two methods in detectingarc and straight line. On this basis, this paper puts forward an improvedRandomized Hough transform, which judges whether the edge point sets generatedfrom random sampling are on the same based on the information of image edgegradient direction, and then uses Randomized Hough transform for further detection,so that we can avoid low detection caused by a lot of invalid accumulation. Thisexperiment detects parameter for2d and3d image by using the algorithm.What ismore, the result is satisfying, which confirmed that the proposed method is of highaccuracy, repeatability, and will play a role in the detection of the space industryparts. If a large number of aviation parts CCD image are deal with one by one inorder to obtain them, it will take much time and effort. So this paper classify theaerospace parts into five categories basing on the research of the CCD imageaviation parts. At the same time, it combines the technology of high resolution CCDimage preprocessing with the Hough transform theory, and develops a kind ofpackage which can check the geometric parameters of the2d and3d imagecontour.