A fast template matching method with rotation invariance by combining the circular projection transform process and bounded partial correlation

This letter proposes an algorithm for rotation invariance template matching by combining the circular projection transformation (CPT) and the bounded partial correlation (BPC). CPT is a method that involves rotation invariance for template matching, but calculating the CPT values from the spatial domain is time-consuming. We use the frequency domain approach to reduce the computation time of CPT and the BPC to rapidly skip the positions that dissimilate to a template image. Experiments demonstrated that the proposed algorithm can considerably accelerate the template matching process when scene rotations are present.