Front-End Vision: A Multiscale Geometry Engine

The paper is a short tutorial on the multiscale differential geometric possibilities of the front-end visual receptive fields, modeled by Gaussian derivative kernels. The paper is written in, and interactive through the use of Mathematica 4, so each statement can be run and modified by the reader on images of choice. The notion of multiscale invariant feature detection is presented in detail, with examples of second, third and fourth order of differentiation.

• Type: Conference
• Year: 2000
• Where: BMCV
• Authors: Bart M. ter Haar Romeny, Luc Florack
• Keywords: BMCV 2000, Computer Vision, Gaussian Derivative Kernels, Multiscale Invariant Feature, Visual Receptive Fields