Analysis of video sequences for anomaly detection using block based approaches

Abstract

Presently it is very important both in private and public environments to monitor activities in Video
surveillance applications . In this context, this paper presents a novel block-based approaches to detect abnormal event
situations by analyzing the pixel-wise motion context. We proceed using motion estimation techniques to characterize the
events at the pixel level. Optical flow is used to extract information such as density and velocity of motion. The proposed
approaches identifies abnormal motion variations in regions of motion activity based on the entropy of Discrete Wavelet
Transform and Discrete Cosine Transform coefficients. We will report successful results on the detection of abnormal
events in video datasets.