Design and realization of a new correlation tracker algorithm

A two-stage image matching algorithm was developed, in which coarse searchi ng and accurate-location are combined. Taking the advantages of stability o f the gray level and other informatin of the object's center , and the rela tive invariance of the series images and the self-relativity of the object, this method provides a new creditability evaluation method. This new match ing algorithm performs perfectly in the condition of complicated background and false objects and high noise, certain rotation and size variance, and is very fast at the same time. Finally, this paper described the realizatio n method of the matching algorithm by use of CPLD and DSP.