Wavelets, vision and the statistics of natural scenes

Abstract

Abstract Over the last decade, the wavelet transform in its various incarnations has grown to be a highly popular means of analysis, with a wide range of applications in processing natural signals. Although there is some debate regarding who developed the first wavelet transform, most of the claims of priority apply to only this century. In this paper, we consider wavelet-like transforms that pre-date these recent studies by possibly as much as several hundred million years. These wavelet-like transforms are found within the sensory systems of most vertebrates and probably a number of invertebrates. The most widely studied of these is the mammalian visual system. This paper focuses on recent work exploring the visual system’s response to spatial patterns and on recent theories of ‘why’ the visual system would use this strategy for coding its visual environment.