Action recognition using silhouette sequences and shape descriptors

Abstract

The paper provides an approach for human action recognition based on shape analysis. The developed approach is intended for specific type of data, namely sequences of binary silhouettes representing a person performing an action, and consists of several processing steps including shape description as well as similarity or dissimilarity estimation. The approach can deal with sequences of different length without removing any frames. The paper also provides some experimental results showing the classification accuracy and overall recognition effectiveness of the proposed approach using several popular shape description algorithms, namely the Two-Dimensional Fourier Descriptor, Generic Fourier Descriptor, Point Distance Histogram and UNL-Fourier Descriptor.