Optical flow cannot be computed locally, since only one independent measurement is available from the image sequence at a point, while the flow velocity has two components. A second constraint is needed. A method for finding the optical flow pattern is presented which assumes that the apparent velocity of the brightness pattern varies smoothly almost everywhere in the image. An iterative implementation is shown which successfully computes the optical flow for a number of synthetic image sequences. The algorithm is robust in that it can handle image sequences that are quantized rather coarsely in space and time. It is also insensitive to quantization of brightness levels and additive noise. Examples are included where the assumption of smoothness is violated at singular points or along lines in the image.

Abstract not found

This paper presents a new hybrid coding method for low bit-rate videoconferencing transmission over ISDN. The use of both predictive and transform coding techniques along with motion compensation allows to obtain bit-rates suitable for ISDN applications. To that extend, a new quad tree based segmentation technique to detect stationary and moving parts of an image is introduced. On the other hand, to fully exploit the statistical properties of the signal to be encoded, an adaptive predictor is designed as well as a transform coding scheme to encode the prediction error. A buffer design that provides constant bit-rate is also introduced. Good visual results are obtained in the range of m x 64 Kbits/s, m = 1,2,3,... . 1. Introduction A video conferencing system is defined as a system that provides a mean to transmit moving images in real time from participants in a conference or in a meeting. The single person-camara (SPC) system is the most commonly used video conferencing system 1 . Thi...

There is one method, called the Block Gradient Method here, whereby to detect image movements from an image sequence. This method has an advantage over the computing costs in a movement vector. However, this method has a general problem, in that the reliability of the detected move-ments varies with varying conditions, such as the image texture. Therefore, it is important to obtain some indices which indicate the reliability of the detected movements for sequential tasks which use them. In this paper, the reliabil-ity of the detected movement by this method is discussed and indices indicating it are proposed. The effectiveness of the indices is shown by experiments. A movement vector stabilizing process, using the indices, is also described. The algorithm proposed in this paper will offer the suitable movement information for the information integration, which is processed in the successional tasks.