Accelerating Template Matching for Efficient Object Tracking

Template matching is used to determine the degree of similarity between two images of the same size. Pearson's Correlation Coefficient is applied, due to its property of invariance to brightness changes. This coefficient is computed for each image pixel, entailing a computationally intensive task. In order to accelerate this process, a dedicated co-processor was designed to implement this computation. To improve the search for the maximum correlation point between the image and the template, we used, in this work, Bacteria Foraging Optimization, one of the swarm intelligence strategies. The search process is run by an embedded general purpose processor. The work presented in this paper describes the implementation of the embedded system and compares the results obtained here to those previously obtained when using other swarm intelligent strategies.