Optic-flow information extraction with directional Gaussian-derivatives

This work is intended to give some ideas to extract motion information from an image sequence. A directional energy is defined in terms of the 1D Hermite transform coefficients of local projections. Each projection is described by the Hermite transform, resulting in a directional derivative analysis of the input at a given scale. Gaussian-derivative operators have long been used in computer vision for feature extraction and are relevant in visual system modeling. We demonstrate that the Hermite transform coefficients of local projections are readily computed as a linear mapping of the 3D Hermite transform coefficients through some projecting functions. The directional response is used to detect spatiotemporal patterns that are 1D or 2D. Practical consideration and experimental results are also of concern