Edge-based compression of cartoon-like images with homogeneous diffusion
Pattern Recognition
Hi-index | 0.00 |
In the computational theory of human vision, developed by D. Marr and his school, zero-crossings detection is proposed as one of the first stages of human visual information processing. The justification of this theory reguires that the zero crossings are rich in information about the original image. In a paper, T. Poggio, H.K. Nishihara and K.R.K. Nielsen stated that a successful extension of B. Logan's analysis to two-dimensional images represents one of the critical unsolved steps of this computational theory. Moreover, such an extension is crucial to the question of spatiotemporal interpolation in vision. In this paper, we shall provide a solution to this problem using a theory of spherical perspective and spherical harmonics.