In this paper

the uniqueness of image matching location using projection feature is proved. The computational method of projection feature using Walsh functions as kernel is presented and the computational process is speeded up. By using linear model established between relative displacement of window image and projection sequence

the fast mapping location is performed. Two combinatory fine location approaches are presented. A large number of experimental results confirmthat the location accuracy of the approaches is approximate to that of SSDA

but the speed is increased more than ten times over that of SSDA.