A size and position invariant event-based human posture recognition algorithm

Abstract

In this paper we report a size and position invariant human posture recognition algorithm. The algorithm employs a simplified line segment Hausdorff distance classification and uses projection histograms to achieve size and position invariance. Compared to other existing method utilizing line segment Hausdorff distance, the proposed algorithm reduces the computation complexity by 36000 times, for our test images. Combining bio-inspired event-based image acquisition and hardware friendly feature extraction and classification algorithm will lead to a promising technology for use in wireless sensor network.