COMPARATIVE-ANALYSIS OF WAVELET MODELS AND VISION

In this paper, we compare the visual stimulus decomposition with the wavelet multiresolution model. The retinal image is analyzed by a multifrequency channel decomposition and then the information is transmitted to the visual cortex. We describe a multiresolution decomposition of the stimulus into a wavelet base. The chosen stimulus is a continuous luminance variation between two fields of respective luminances L1 and L2. The variation is described by the function : B (b, x) = 1/2(tanh (x/b) + 1). The wavelet analysis of the function B shows that the wavelet transform decreases (absolute values) with area width. With a sharp area, the transmitted information is more important than that corresponding to a wide area.