A neural network based motion-estimation technique is developed, that is applicable to sub-pixel as well as large movements. Experiments on typical test frame sequences indicate marked improvement in accuracy of motion vector estimates over the MPEG logarithmic block matching algorithm. The method utilizes a modified Hopfield neural network. Due to the neural network's fault-tolerant nature and parallel computation capability, fast, accurate, and reliable results are obtained. Application to MPEG based video compression is also discussed.<>
Neural network based motion vector computation and application to MPEG coding
Proceedings of 1st International Conference on Image Processing ; 2 ; 938-942 vol.2
1994-01-01
496887 byte
Conference paper
Electronic Resource
English
Neural Network Based Motion Vector Computation and Application to MPEG Coding
British Library Conference Proceedings | 1994
|Design of adaptive quantizer for motion picture experts group (MPEG) video coding [2451-34]
British Library Conference Proceedings | 1995
|Motion Object Detection Based on MPEG-4 Video
British Library Online Contents | 2008
|Motion vector prediction in multiview video coding
IEEE | 2005
|PSO Based Bit Rate Optimization for MPEG-1/2 Video Coding
British Library Conference Proceedings | 2005
|