THE HERMITE TRANSFORM-THEORY

A digital image is usually specified by an array of pointwise intensities. For an intelligent interpretation of the image data we need to make the important information explicit. This usually implies determining spatiotemporal relationships between these intensities, and hence requires some form of local processing of the visual data. In this paper we introduce a new scheme for the local processing of visual information, called the Hermite transform. We have addressed the problem from the point of view of image coding, and therefore the proposed scheme is presented as an analysis/resynthesis system. Our objectives are, however, not restricted to coding. First, the analysis part is designed so that it can also serve applications in the area of computer vision. Indeed, derivatives of Gaussians, which have found widespread application in feature detection over the past few years, play a central role in the Hermite analysis. Second, next to integrating ideas from the distinct, but related, areas of image coding and computer vision, it is argued that the proposed processing scheme is also in close agreement with our current insight into the image processing that is carried out by the human visual system. © 1990 IEEE