The adaptive modulated wavelet transform image representation

In this paper, a new, called the adaptive modulated wavelet transform (AMWT) image representation is presented. One of the attractive features is that the informative instantaneous frequencies of images can be taken into account to improve the representation performance via adaptation of the modulating frequencies involved. The transform coefficients in both wavelet and modulated wavelet domains are uniformly quantized with several quantization levels. The computed peak signal-to-noise ratio values and entropies are used as rate distortion curves for performance comparison. Experimental results show that AMWT out performs wavelet transform for representing images containing textures with rapid variation in grays.