Jb. Luo et al., A SCENE ADAPTIVE AND SIGNAL ADAPTIVE QUANTIZATION FOR SUBBAND IMAGE AND VIDEO COMPRESSION USING WAVELETS, IEEE transactions on circuits and systems for video technology, 7(2), 1997, pp. 343-357
Discrete wavelet transform (DWT) provides an advantageous framework of
multiresolution space-frequency representation with promising applica
tions in image processing, The challenge as well as the opportunity in
wavelet-based compression is to exploit the characteristics of the su
bband coefficients with respect to both spectral and spatial localitie
s, A common problem with many existing quantization methods is that th
e inherent image structures are severely distorted with coarse quantiz
ation, Observation shows that subband coefficients with the same magni
tude generally do not have the same perceptual importance; this depend
s on whether or not they belong to clustered scene structures, We prop
ose in this paper a novel scene adaptive and signal adaptive quantizat
ion scheme capable of exploiting both the spectral and spatial localiz
ation properties resulting from wavelet transform, The proposed quanti
zation is implemented as a maximum a posteriori probability (MAP) esti
mation-based clustering process in which subband coefficients are quan
tized to their cluster means, subject to local spatial constraints, Th
e intensity distribution of each cluster within a subband is modeled b
y an optimal Laplacian source to achieve the signal adaptivity, while
spatial constraints are enforced by appropriate Gibbs random fields (G
RF) to achieve the scene adaptivity, Consequently, with spatially isol
ated coefficients removed and clustered coefficients retained at the s
ame time, the available bits are allocated to visually important scene
structures so that the information loss is least perceptible. Further
more, the reconstruction noise in the decompressed image can be suppre
ssed using another GRF-based enhancement algorithm, Experimental resul
ts have shown the potentials of this quantization scheme for low bit-r
ate image and video compression.