Title :
Embedded Region-Based Image Coding using Integer Wavelet and Single Quadtree Partitioning
Author_Institution :
Beijing Normal Univ., Beijing
Abstract :
Region of Interest (ROI) coding belong to the efficient region-based image compression method and is an important feature of JPEG2000 standard. In this paper, a new ROI coding method is proposed. In the method, we utilize the improved SPECK algorithm instead of EBCOT coding recommended by JPEG2000 to encode all scaled coefficients of ROI and BG. Because coefficient clustering and single quadtree partitioning scheme is applied, the presented method has the low computational complexity. Experimental results show that the new method cannot only efficiently encode these bitplanes shifted by Maxshift method or the general scaling based method, but also obtains lower algorithm complexity than these methods using EBCOT in JPEG2000.
Keywords :
computational complexity; data compression; image coding; quadtrees; wavelet transforms; EBCOT methods; JPEG2000 standard; Maxshift method; SPECK algorithm; coefficient clustering; computational complexity; embedded region-based image coding; integer wavelet; quadtree partitioning scheme; region of interest; Clustering algorithms; Cybernetics; Educational institutions; Image coding; Image storage; Machine learning; Partitioning algorithms; Propagation losses; Transform coding; Wavelet transforms; EBCOT; Eegion of interest; Image coding; JPEG2000; Wavelet transform;
Conference_Titel :
Machine Learning and Cybernetics, 2007 International Conference on
Print_ISBN :
978-1-4244-0973-0
Electronic_ISBN :
978-1-4244-0973-0
DOI :
10.1109/ICMLC.2007.4370169