Sub-pixel point detection algorithm for point tracking with low-power wearable camera systems: A simplified linear interpolation

Abstract

With the continuous developments in vision sensor technology, highly miniaturized low-power and wearable vision sensing is becoming a reality. Several wearable vision applications exist which involve point tracking. The ability to efficiently detect points at a sub-pixel level can be beneficial, as the accuracy of point detection is no longer limited to the resolution of the vision sensor. In this work, we propose a novel Simplified Linear Interpolation (SLI) algorithm that achieves high computational efficiency, which outperforms existing algorithms in terms of the accuracy under certain conditions. We present the principles underlying our algorithm and evaluate it in a series of test scenarios. Its performance is finally compared to similar algorithms currently available in the literature.