Match The Outline Of The Product Tomographic Images

In reverse engineering, the contour recognition and matching of the product slice image is one of the key links in the CAD modeling technology based on sectional image, at the same time it抯 an important problem in the field of graph and image processing. Aimed at the reverse problem of complex mechanical products containing branches or trapped volumes, the author carried out the study for the planar contour recognition and slice contour matching in this paper. First, the including relationship of contours on one plane is mapped into a contour tree, with which the in-out feature of the contours on one plane is recognized. Next according to the slice sequence, these contour trees constitute a contour forest, which offers the restraint mechanism for the contour matching. Then with the applying of the ellipse-fitting algorithm, the slice contour matching is realized. Having studied the technology of intelligent pattern recognition, the author also brings forward amatching algorithm based on artificial neural network, which optimizes the traditional algorithm. At last, the programming to verify the two algorithms and the comparison between them is introduced in the paper. The detail research works are as follows: U Setting up the contour forest and realizing the recognition of the in-out feature on one plane. The author maps the including relationship of contours on one plane into a contour tree. On the basis of the crosswise restrict from the contour tree, the in-out feature of the contours on the plane is recognized. At the same time, the contour trees are joined to a contour forest that provides the lengthways restrict for the following slice contour matching. ?Appling the classical ellipse-fitting algorithm to realize the slice contour matching. Based on the contour forest, the author makes use of the five parameters of ellipse to describe the shape and location information of the contours. Then according to their comparability and linear variety, the slice contours are grouped and matched. ?Putting forward the optimization algorithm for slice matching based on the artificial neural network. Aimed at the difficult matching problem caused by noise, flaw, distortion or high complexity of the product, the author applied the technology of ANN and intelligent pattern recognition into the matching algorithm. That is the contours are regarded as different patterns, and a tn-layer BP network is designed to recognize them. With the recognition, the contour matching is realized. ?Exploiting a program to analyze and verify the above two algorithms. Under the environment of VC++, the author exploits a program to realize the two algorithms. The OpenGL 3-D graphical interface and the ActiveX component technology are also applied to show the recognition and matching result dynamically. The validity and feasibility of the algorithms is verified in this part, and the comparison between them is also given.