67 巻 (2001) 8 号 p. 1271-1275
This paper proposed a method for image template matching that is robust to the occlusion of a target image area. A template image is separated into local block areas and the normalized correlation rates are calculated in all block areas. Only partial block areas that may be useful for matching are selected according to the thresholds for the normalized correlation rates in all block areas that are set through the template matching using the reference images. The experimental results show that the proposed method was robust to the occlusion of the target image in comparison with the conventional methods because it can search the target image area successfully even if the 90% of the target image area is occluded.