A binocular vision tracking method based on Gabor filter

To solve the stability and rapidity problem of visual tracking, the paper presents a method that the characteristic points in sequenced images assemblies for target is collected by the Gabor filter, and turned into matching templates. The detection with the Sobel operator is used to get initial points position, to determine each initial point parameter, and in this way we can speed up optimizing for parameters. The experimental result has been verified that the arithmetic is with highly reliability and availability, it can avoid the affection by target feature collecting from background, so the method has the benefit to the real time binocular tracking.