Research On Feature Recognition Of Engineer Vehicle

With the development of society fertility, automatic management becomes a trend, which brings people much convenience in work and daily life and improves the efficiency. The application of automatic management is connected with computer science and technique level tightly. Rapid development and application of computer and technique makes it possible to widely popularize automatic management techniques. And vehicle automatic management is necessary, as the number of vehicles increases continuously. In this application background, the dissertation researches on engineer vehicle recognition system model based on the profile of the vehicle. The work is divided into several aspects as following.Firstly, segmentation algorithm is studied here. This dissertation introduces several color spaces and illuminates their characteristics. Segmentation of vehicle is conducted in HSV color space. With Euclid distance computed in HSV space, color image is translated into grayscale image, which segmentation methods can be applied in to partition the vehicle from the background. An algorithm, which combines SEM algorithm and region growing method, is put up to segment target. And the results of experiments on many images imply the efficiency of our method.Secondly, feature extraction methods based on target profile is researched. For segmented target, Fourier descriptor is adopted. Than principal component analysis is used to extract profile information.Finally, methods of vehicle shade elimination are discussed here. The dissertation does research on shade elimination methods deeply. Two methods are put up. And experiment results show that both of these methods have some performance.