Fast and Accurate Template Matching Using Pixel Rearrangement on the GPU

Abstract
A GPU (Graphics Processing Unit) is a specialized processor for graphics processing. GPUs have the ability to perform high-speed parallel processing using its many processing cores. To utilize the powerful computing ability, GPUs are widely used for general purpose processing. The main contribution of this paper is to show a new template matching algorithm using pixel rearrangement. Template Matching is a technique for finding small parts of an image which match a template image. The feature of our proposed algorithm is that using pixel rearrangement, multiple low-resolution images are generated and template matching for the low-resolution images is performed to reduce the computing time. Also, we implemented our algorithm on a GPU system. The experimental results show that, for an input image with size of 4096 × 4096 and a template image with size of 256 × 256, our implementation can achieve a speedup factor of approximately 78 times over the conventional sequential implementation.

This publication has 14 references indexed in Scilit: