Improvement Of Industrial CT Image Vectorization System Towards Reverse Engineering

The model reconstruction with complex free surface in Reverse Engineering canoffered by Industrial Computerized Tomography. However, the2-dimensional grayslicing image which is obtained by the ICT system have to be executed vectorization forusing it in CAD software. The developed industrial CT system is low efficiency, poorstability and low accuracy, which affect reconstruction. So the improvement researchwas realized in this paper:Image pre-processing methods were studied, including image enhancement,threshold segmentation and edge detection operator. Median filter technology was choseas image enhancement method through comparison between median filter andhistogram equalization. According to different industrial CT image, iterative thresholdmethod and adaptive threshold were both chose as image threshold segmentation.Improved Canny edge detection algorithm was chose as edge detection method throughcomparison among the typical edge detection algorithms.Feature point detection was researched, one method was proposed which metpositioning requirement of industrial CT image. Firstly, candidate feature point wasobtained based on neighborhood sorting; Secondly, a lot of redundant feature pointswere removed by curvature algorithm.An improved point Hough transform circle detection method was proposed basedon circle detection methods were studied. Firstly, candidate point groups were obtainedby two times filter. Secondly, circle detection was executed by adaptive step value andaccumulator judging value. Erroneous detection and leak detection were both avoided.At the same time, the precision of circle detection and fitting was improved as well.At last, the improved result was turned out to be validated by comparing withoriginal system.