Object Tracking Based On Gabor Wavelet

The moving target's tracking are applied widely in the both civil and military field. In this dissertation, we adopt the Gabor wavelet to abstract the characters of the moving target's sequence image to compose the matching template to realize the moving target's tracking.We adopt the Gabor wavelet to abstract the target's characters for avoiding background effect. By using the LM arithmetic to optimize the wavelet's parameter and using Sobel operator to detect the initial points, we can ascertain the initial point's kinds of parameters, and improve the parameter optimization's speed.In the process of tracking, we use LM optimization arithmetic to abstract the target's Gabor wavelet feature template and track the object. We use LM optimization arithmetic again to make the wavelet character template changing along with the target's rotation, plane-moving, scale and other affine transforms, so we can obtain the target's affine moving parameters to realize stable tracking for the target.When the changing of the target stance is too tempestuous to make some characters disappear from the image, we use Greedy algorithm to update the position of the Gabor wavelet characters to realize the character template's adaptive transmogrification, and the template's updating ascertains the tracking stability. At last, we use experimentations to validate the reliability and impact of the algorithmic.