L. Zhang, H. Miike and K. Kuriyama
Graduate School of Science and Engineering, Yamaguchi University, 2557 Tokiwadai, Ube 755, Japan
(Received November 2, 1998; Accepted December 21, 1998)
Keywords: Optical Flow, Global Optimization, Local Optimization, Spatio-Temporal Optimization, Gradient-Based Technique
Abstract. The gradient-based technique is one of the popular methods to determine optical flow (apparent two-dimensional velocity field). It can be classified into two broad categories: the global optimization and the local optimization. Horn and Schunck proposed the spatial global optimization to constrain the estimated velocity field. In this study, a new approach is proposed to extract the reliable optical flow fields. This is the gradient-based spatio-temporal optimization with combination of local and global approach. Main point is the introduction of a spatio-temporal measure (locally averaged) of the error function in the conservation equation. Experiments have been made to confirm the performance of the proposed method and to clarify the difference of characteristics between them.