02-08-2011, 04:39 PM
Presented By
V. RAJITHA P.KAUMUDI
[attachment=15008]
ABSTRACT:
In this paper, we propose a simple and efficient technique for embedded color image coding. The proposed algorithm exploits inter- and intra-sub band correlations of the wavelet transformed luminance (Y) and chrominance (U and V) planes as well as the interdependency among the coefficients of the three-color planes. The coefficients of the three-color planes are linked through the composite spatial orientation trees (CSOT) having root nodes in the Y-plane only. The CSOT is defined for blocks of coefficients rather than a single coefficient. Thus, the proposed algorithm combines the features of both zero-tree and zero-block based algorithm into a single algorithm due to the use of block-tree hierarchical structure. Simulation results show the improved performance of the proposed algorithm over other sate-of-the-art coders such as MPEG-4, color SPIHT (CSPIHT), color SPECK (CPECK), and JPEG-2000.
I. INTRODUCTION
In recent years, wavelet based scalable image coding techniques [4]-[1]-[4] have attracted increased interest due to their excellent rate distortion performance, low encoding and decoding complexity, and rate scalability. The wavelet transformed images have both inter - and intra - sub band correlations among the coefficients. Most of the wavelet based coding techniques use either zero-tree or zero-block to exploit these correlations. The zero-tree coding algorithms such as EZW [1] and SPIHT [2], exploit inter–sub band correlations among the coefficients across the sub bands only. On the other hand, zero-block coding algorithms like EBCOT [3] and SPECK [4] exploit intra - sub band correlations among the coefficients with in the sub bands only.
The above algorithms though primarily designed for coding of gray-scale images, but have also been extended for coding of color images and/or videos [4]-[6].. The trivial approach for coding of color images is to apply these algorithms independently on RGB or YUV color planes, and then serially concatenating the bit stream of each color plane. However, this would require proper bit allocation between each of the color planes and would not be fully embedded. The SPECK and SPIHT algorithms are extended to YUV color planes in [4] and [6] respectively to generate fully embedded bit streams. In both of these schemes, the list of insignificant pixels (LIP) and the list of insignificant sets (LIS) are initialized with the appropriate coordinates of the top-level sub bands of the three-color planes serially one after the other. The appropriate algorithm (SPECK or SPIHT) is then used to code the three-color planes independently. However, it is a well-established fact that the three-color planes may be uncorrelated, but they are not independent. A color image coding algorithm exploiting the interdependency among color planes was first proposed in [5].It uses EZW type of spatial orientation trees (SOT) with each luminance node having six children, four in the luminance plane and one in each of the two - chrominance planes. However, it suffers with the problem of early accumulation of chrominance nodes in LIP thereby reducing its coding efficiency. An efficient and embedded color image coding technique using SPIHT's SOT is proposed in [7], in which all three-color planes are linked through CSOT with root nodes in luminance plane only, but a root node has all of its descendents either in the luminance plane or chrominance planes. Kassim and Lee [8] also used the same SOT structure for 4:4:4 YUV color formats. These techniques efficiently exploit the interdependency of three color planes and similarities of wavelet coefficients across the sub bands only. However, in order to further increase the coding efficiency, there is need to exploit the intra-sub band correlations of wavelet coefficients as well.
