Digital signal processing in video surveillance

Ugwuanyi, Stephen Okwudili

This paper presents two principles of Digital Signal Processing (DSP) techniques in Video Surveillance technology. First, noise due to camera and quantisation process estimation which has been a persistent problem in video surveillance systems was evaluated using variance method of noise estimation. Second was the evaluation of the effects of bitrate on Peak Signal to Noise Ratio (PSNR) using search range and slice of increasing values. The simulation which was carried out using H.264/AVC-16.1 shows that small values of search range and slice produced good results for both bitrate and PSNR. Implying that search range and slice have exponential rise relationship with the average frame in motion (vector). Unlike other traditional motion estimation techniques, like H.26x, H.264/AVC is an efficient video compressor using bitrate with less computational complexity and good image quality.

Download PDF: