Fast requantization using self organizing feature map with orthogonal polynomials transform

In this paper, a fast requantization algorithm for codebook generation using Self Organizing Feature Map (SOFM) with Orthogonal Polynomials Transform (OPT) is proposed for color image encoding. The proposed requantization algorithm is implemented in transform domain and the transformed training vectors are obtained with Orthogonal Polynomials Transformation in order to construct the quantization codebook Q1. Since the OPT concentrates most of the energy within the transformed training vector into low frequency coefficients, half of the high frequency coefficients are discarded so as to achieve efficient bit-rate reduction and hence reduces the codebook construction time. The quantization codebook is constructed from truncated transformed training vectors and the proposed requantization codebook is obtained from codebook Q1. The SOFM algorithm is employed in both quantization and requantization stage in order to speed up the codebook generation process. The computation time of codebook construction for color image is further reduced in this proposed work by constructing a single codebook for all the R, G and B color components, utilizing the inter-correlation property of the OPT coding. The experimental result shows that the codebook construction time of the proposed requantization algorithm for color image encoding is considerably reduced when compared with the existing Chih-Tang Chang et. al. requantization[3].