A new method of two-stage image matching with multiple assumption tests based on two-fork deciding tree is proposed in this paper. First

the definition of twofork deciding tree representing multiple assumption tests and its computational cost formula are given. Then

a new variable threshold is derived. Finally

the computational cost of two-stage image matching with multiple assumption tests using variable threshold is obtained. A large amount of computer simulations shows that the limit of computational cost of original twostage template matching can be broken by the proposed method and the computational cost of image matching is further reduced. Furthermore

it is sure that the matching location accuracy is almost the same as the Mean Absolute Difference Algorithm.