Error analysis of DPCM video signal processing

An error analysis of the 2D FIR digital filters realisation using DPCM video signal processing is outlined. Quantisation errors due to coding and finite word length are studied. Simulation results demonstrating the coherence between the theoretical and the experimental results are presented. The main conclusion that can be drawn from the theoretical and empirical results is that there is a minimum value of the filter coefficients´ word length after which no improvement in the signal-to-noise ratio (SNR) is possible. However, subjective tests have shown that images obtained using the DPCM realisation with this minimum word length have similar visual quality as that obtained using conventional filters with infinite precision.