Enhancements in the dual tree discrete wavelet transform algorithm for video processing

Abstract

This paper proposes two enhancements to the Noise Shaping (NS) algorithm with the intent of reducing the processing time required to shape the coefficients in the dual tree discrete wavelet transform (DDWT). First, Subband Significant Coefficient Determination (SSCD) will identify the most important coefficients while eliminating the others. A second algorithm, Energy Distribution Noise Shaping (EDNS) more efficiently processes the wavelet coefficients of the transform based on individual subband energy distributions.