Novel template matching method with sub-pixel accuracy based on correlation and Fourier-Mellin transform

Abstract

Template matching is the process of determining the presence and the location of a reference image or an object inside a scene image under analysis by a spatial cross-correlation. Conventional cross-correlation type algorithms are computationally expensive. Furthermore, when the object in the image is rotated, the conventional algorithms cannot be used for practical purposes. An algorithm for a rotation-invariant template matching with subpixel accuracy is proposed based on the combination of the correlation and Fourier-Mellin transformation when the fluctuating scope of the rotation angle is [−20deg,20deg]. The algorithm consists of two stages. In the first stage, the matching candidates are selected using a computationally low-cost improved correlation algorithm. The operation of AND is adopted to reduce the computational cost for this stage. In the second stage, rotation invariant template matching is performed only on the matching candidates using the cross-correlation algorithm after adjusting image with a Fourier-Mellin invariant (FMI) descriptor, and the matching precision is subpixel by the novel method using the Fermat point. Experimental results show that the proposed method is very robust to Gaussian noise and rotation, and it also achieves high matching accuracy and matching precision.