A New Spatiotemporal Approach for Image Analysis. Application to Motion Detection
Abstract
Image sequence analysis involves 3D data. Consequently, we propose a new spatiotemporal global approach for image sequence processing where an image sequence is regarded as a 3D data flow. This approach is illustrated in the case of motion detection in a Markovian framework. This leads to the development of a 3D Markov Random Field based algorithm which takes into account in the same way spatial and temporal dimensions. The required relaxation algorithm runs on (x, y, t) for the whole image sequence. Motion detection results illustrate the efficiency of this algorithm. \textcopyright Springer-Verlag Berlin Heidelberg 1995.