Performance evaluation of progressive image transmission techniques

Progressive image transmission is receiving attention for application in interactive image communication over low-bandwidth channels, In progressive image transmission (PIT), a low resolution image is first transmitted, this is later progressively improved by further transmissions until the received image is reasonably a good representation of the original image. In this paper, we present a comparative performance evaluation of four PIT techniques based on vector quantization (VQ) namely: PIT by residual error VQ in spatial domain (RVQ-SPA), and in transform domain (RVQ-TRA), multistage VQ by optimum bit allocation (MVQ-OBA), and VQ on images in quadtree form (VQ-QT). In VQSPA, VQ is applied on the residual errors in spatial domain. The residual error image is generated by taking the difference between the original image and coded image. A fraction of the image information is transmitted, which is utilized in the reconstruction of the image. In the next technique, VQ is applied on the residua] errors in the transform domain. In MVQ-OBA, VQ is applied on the transform coefficients in sach a way that the total number of bits allocated to each coefficient is proportional to the coefficient variance-, In the last technique the image to be transmitted is first decomposed into a quadtree data structure, VQ is then applied to the difference quadtree (VQ-QT). Our simulation results demonstrate that all the four techniques, exhibit very good coding performance at low bit rates. Here, PIT using discrete cosine transform (OCT) is used as the baseline.