EZW Coding with Improved Execution Time on the Basis of Descendant Scanning of Zero-tree Roots

Abstract

Compression is a very important aspect of maintaining database systems. Thus efficiency of compression systems is important. In todays world the transfer of data too is done in a compressed manner. The bit stream is optimized. One such way is embedded transmission scheme in which the compressed low bit rate information is transmitted first and then the higher rates follow. In this scheme even if some part of bit stream is not received, information corresponding to lower bit rate can be decoded. Thus in such transmission schemes both transmitting and encoding time are important. Embedded Zero-tree Wavelet (EZW) encoding is one such technique for embedded transmission and compression of images. EZW for images has been used to form a video coder also. The main objective of this thesis is to implement and improve EZW coding. The thesis proposes a coding scheme based on EZW, with improved execution time and compression ratio using some constraints. The improvement in the present coding is done by assuming that most of the coefficients values lying in the decomposition subband are low and near to zero. Thus they need not to be checked again and again for significance. This saves encoding time in the proposed coding scheme, based on EZW. The proposed coding scheme reduces the execution time by 2 sec. and compression ratio almost by 2% in 7 iterative passes without any degeneracy in the decoded image as compared with standard EZW coding. Improvement of 1 sec. more can be made in execution of 8 successive passes, if a fall of around 1 dB is allowed in the decoded image of proposed coding as compared to the decoded image of EZW coding at the 8th pass. The first part of this thesis is dedicated in knowing discrete wavelet transform which is the base of the present coding and then the EZW coding itself is covered in detail. In next part the theory to improve the present coding is stated and then the improved coding scheme is proposed. Later, results are brought out on various images in support of the theory and constraints are defined. Finally, future aspects of research with respect to proposed coding scheme are introspected.