Forward adaptive Laplacian source coding based on restricted quantization

A novel solution for Laplacian source coding based on the three-level restricted quantization is proposed in this paper. The restricted quantization provides the reduction of granular distortion with the proper choice of the support region. We use the combination of two three-level restricted quantizers having unequal support regions, which are selected based on the lower distortion. The quantizers are designed using the Lloyd-Max’s algorithm, by assuming the restricted Laplacian distribution of the input signal. The outputs are encoded using the Huffman code. In order to improve the performance the forward adaptive algorithm was employed, where the adaptation to the signal variance (power) was performed on frame-by-frame basis. Theoretical analysis has shown that in this manner the robustness and adaptability of the proposed solution is enabled. The experimental results prove that the proposed switched three-level restricted quantizer is superior in comparison to the three-level unrestricted quantizer, and outperforms the one-bit (two-level) Lloyd-Max’s quantizer, while offering performance comparable to the two-bit (four-level) Lloyd-Max’s baseline with large savings in bit rate.