Co-design system for template matching using dedicated co-processor and modified elephant herding optimization

Abstract

Template matching is an important technique used for object tracking, to find a given pattern within a frame sequence. Pearson's Correlation Coefficient (PCC) is widely used to quantify the degree of similarity of two images, because its properties of invariance to brightness changes. This coefficient is computed for each image pixel, and it entails a computationally very expensive process. In this paper, aiming at accelerating this process, we propose to implement the template matching as an embedded co-design system. In order to reduce the processing time, a dedicated co-processor, which is responsible for performing the PCC computation is designed and implemented. Elephant Herding Optimization (EHO) is used to improve the search for the maximum correlation point of the image and the used template. The search process is implemented in software and is run by an embedded general purpose processor. The performance results are compared to PSO's and Exhaustive Search's (ES) performance. The designed system achieves processing time and success rate requirements as needed in real-time applications.