The Research Of Feature Extraction And Tracking Method Based On Gabor Wavelet

Target tracking is one of core technologies in computer vision, and has the extremely vital significance and the widespread practical value. Feature point extraction and tracking method is an important contents to study imaging-based target tracking, and avails to the research on target identification, maneuver detection and so on.In this paper, Feature point extraction and tracking method based on image-based tracking technologies are researched. After deeply studying the feature of the Gabor transformation, a feature point extraction method is proposed, which is based on Gabor transformation, edge detection and energy function. The vectors of the feature points extracted are used as the characteristic vectors, and point pattern tracking and trace forecasting are both introduced. In order to get higher performance, A tracking method based on feature points is researched.The main achievements are summarized as follows:1. Firstly image-based tracking technologies and their significances are introduced. And then Methods and status quo of target tracking are concluded, Gabor wavelet feature point extraction is also simply mentioned.2. Target feature extraction based on 2D Gabor wavelet transformation. After deeply studying the feature of the Gabor transformation, a characteristic vector for describing points feature is defined to realize feature point extraction. Comparing with gray template matching method, wavelet feature template is the feature on target itself, not all the pixels of tracking window, which avoids background pixels, restrains the disturb of background. The feature points extracted in image edge points make all these points converge in target feature.3. Point pattern tracking concept is employed with a matching method based on Gabor wavelet. This arithmetic is improved by Nearest Neighbor Search Algorithm, and finally find the matching pair in the matching set. 4. In order to forecast the target position in the next frame, kalman filter method is researched. The simulation results verify that this method employed in the paper is effective.