The statistical properties of image subbands are studied for various subband decompositions. The authors observe that the amount of energy in image subbands increases towards the lower frequency subbands and, more importantly, towards the image edges, simultaneously. They also find that the directionality of energy distribution is highly dependent on the orientation of the edges. Based on the energy distribution, they extract arbitrarily shaped regions in each subband and employ the entropy-constrained quantizers using the generalized Gaussian distribution for modeling the image subbands. They also address the problem of determining an optimal subband decomposition among all the possible decompositions. The resulting feature based subband coding scheme yields at least 1-dB better PSNR than previously proposed schemes for the "Lena" image.<>
Region based subband image coding scheme
Proceedings of 1st International Conference on Image Processing ; 2 ; 859-863 vol.2
1994-01-01
548828 byte
Conference paper
Electronic Resource
English
Region Based Subband Image Coding Scheme
British Library Conference Proceedings | 1994
|An Image Coding Scheme using Block Prediction of the Pyramid Subband Decomposition
British Library Conference Proceedings | 1994
|NTRS | 1994
|Optimal Entropy Coding in Image Subband Coders
British Library Conference Proceedings | 1994
|