GPU-accelerated feature tracking

The motivation of this research is to prove that Graphics Processing Units (GPUs) can provide significant speedup of long-executing image processing algorithms by way of parallelization and data throughput. This research accelerates the Kanade-Lucas-Tomasi (KLT) feature tracking algorithm using OpenCL. KLT can easily suffer from low frame rates when tracking many features in a video sequence. The experiments conducted prove that when tracking over 500 features in a high definition (HD) dataset, GPU-based KLT provides a 92% reduction in total runtime compared to a CPU-based implementation.