Abstract. In this paper, we present novel image-derived, invariant features that accurately capture both the geometric and color properties of an imaged object. These features can ...
This paper introduces a novel statistical mixture model for probabilistic clustering of histogram data and, more generally, for the analysis of discrete co occurrence data. Adoptin...
We apply the concept of network-consciousness to image compression, an approach that does not simply optimize compression, but which optimizes overall performance when compressed i...
ÐWe propose a novel approach for solving the perceptual grouping problem in vision. Rather than focusing on local features and their consistencies in the image data, our approach ...
We propose a new class of mapping-based nonlinear quadratic Volterra filters for image enhancement. The input signal is mapped prior to filtering by a memoryless one-to-one nonl...
Reinhard Bernstein, Sanjit K. Mitra, Michael S. Mo...