Abnormality Detection Using Image Quality Estimation Technique Through Border Surveillance For Wireless Transmission

Presently abnormality or anomaly detection from video streams of surveillance system has drawn quite a significant research interest. Modern surveillance system uses pervasive CCTV for recording surveillance video, so recording or capturing of video is not a problem. The main difficulties lie in the availability of human resources to observe and analyze the video clips for long hours. They are not only very exhaustive but are mostly quite costly. Automatically finding of abnormal or risky behavior significantly decreases the overhead to label and annotate the surveillance videos of huge number of hours. In the proposed technique abnormal changes in a surveillance frame is detected and achieved through image quality estimation by using Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index Metric (SSIM). The proposed anomaly detection is done automatically at the remote place of surveillance where uncorrelated (less similar) frames of the video represents anomalies or abnormality. False alarms are avoided by sending the reference and abnormal frame the control room using 1-D DWT and 2-D DWT at the remote surveillance place. The results reported are quite encouraging and the model will suitably work in any wireless communication domain.