Yan H-S, Tjahjadi T

Proceedings of Joint Conference on Information Sciences, September 2003, 1390-1395.

Abstract

In this paper we concentrate on investigating the first-order differential method for optical flow estimation and propose a novel velocity-based surface fitting technique. The surface fitting not only smoothes the image data, but also uses a polynomial function to represent the intensity of the fitted image. This integrates pre-filtering procedure and the intensity gradient estimation and thus provides a simple approach to optical flow estimation. The velocity-based technique adapts the degree of smoothing by changing the size of the fitting window based on the calculated motion field, and therefore produces a more accurate optical flow. The strengths of our approach are its algorithmic simplicity and its good performance in most synthetic and real image sequences, especially those containing simple texture.