Research On Algorithms Of Segmentation And Contour Feature Of Medical Dental Images

Disastrous accidents like explosion, tsunami, plane crash and car accidents well as criminal cases occur repeatedly make the application requirements of individual identification technology has become more and more urgent. Physiological features which are often used to identify persons like fingerprints, iris and palm prints are unable to save for a long time because of their properties being as organic material, so they will lose their real values to recognition. Although the precision of DNA detection is very high, but the high cost, long testing period and the accuracy affected by the quality and quantity of materials make it Lose practicability. Due to the protection of the jaw and stability of itself, teeth have advantages of non-corrosion and recalcitrance. And the development of oral medicine and the ascension conscious of the protection about tooth make dental images are accumulated and preserved in large quantities. The differences between teeth from individuals make it possible for tooth as the feature for human identification.The technology of human identification is used to decide the identity of persons by judging biological characteristics by the means of the image processing technology. There are two kinds of biological characteristics:physiological feature and behavior characteristics. Characteristics such as fingerprint, iris and palm print all belong to the first kind and voice, gait, the signature belongs to the behavior characteristics. Usually the characteristics selected to identification need to be difficult to counterfeit unique, measurable and stable in a certain period. Teeth segmentation and positioning is one of the key steps in the system of individual identification based on the shape of tooth. The integrity of segmentation directly determines the effectiveness of the identification results. The result of feature extraction will also seriously affect the subsequent operation like classification and similarity measurement.This paper firstly introduces the definition of the technology of digital image segmentation and feature extraction and the significance as also as the development situation at home or abroad. Then we study the method of segmentation and contour feature extraction of dental image (bitewing). The target of dental segmentation is to make single teeth in different areas from the whole image. Watershed algorithm belongs to the field of mathematical morphology is one of the image segmentation algorithms based on areas. We can get one-pixel wide and closed edges. And it also behaves well in responding to the weak edges. But the phenomenon of over-segmentation in the results of traditional watershed algorithm is very serious. Dental images have the some special characters like the difference of the gray value between foreground and background is very small and there are adhesions between adjacent teeth and heavy noises. We can use morphological operations to preprocess the image in order to improve the readability of dental images and eliminate the adhesion phenomenon between objects tooth. Contour as one of the most basic characteristics of image is an essential part of the image analysis. Active Contour model namely the Snake model combines prior knowledge and the characteristics of the image itself has the strong ability of extracting targets from specific areas. In this paper, we propose a method of setting initial contour which combines with the result of the morphology processing in applying the algorithm of GVF SNAKE model to extract teeth contours. It has realized the initial contour automatic set rather than artificial work. Finally, we simulate the proposed algorithm through several dental images and the algorithms we proposed results in accurate segmentations and contour feature extraction.