Video preprocessing techniques for real-time video compression-noise level estimation techniques

Processing of the video signal prior to MPEG2 encoding can provide dramatic quality improvements in the subjective quality of the reconstructed video or bit-rate reductions in the generation of the compressed bit-stream. In a broadcast environment the video source is not guaranteed to be noise free. Noise in video signal will not only reduce video quality but also cost extra bits to code. Instead of using expensive external video pre-processors, state of the art broadcast quality real-time video encoders incorporate onboard video pre-processing to remove noise. An estimate of the noise level is essential for the control of subsequent filtering. Optimum performance is achieved when there exists a close coupling between the video preprocessor and the video encoder. This paper presents noise level estimation techniques in the design of on-board video preprocessor for real-time video encoders. Simulation results are presented along with implementation considerations.