Construction of Three-Level Scalar Quantizer with Extended Huffman Coding for Gaussian Source

Abstract

This paper discusses a three-level scalar quantizer with extended Huffman coding. The signal at the entrance of the three-level scalar quantizer is modeled by Gaussian probability density function. In order to improve the performance of the considered model, the optimal decision threshold depending on the maximum of the signal to quantization noise ratio value is determined in this paper. Performances of the considered quantizer model are estimated by comparing the calculated values of the signal to quantization noise ratio with the three-level scalar quantizer with Huffman coding. Also, detailed analysis of the average bit rate and the source entropy for the proposed model is presented.