A NEW OBJECT MOTION ESTIMATION TECHNIQUE FOR VIDEO IMAGES, BASED ON AGENETIC ALGORITHM

In the search for lower bit rate image compression and representation, a new Video Motion Estimation Technique (VMET), that considers video object translation, as well as rotation, and planar multilayering, is described in this paper. This new concept uses a Modified Multipopulat ion Coevolutionary Genetic Algorithm (MMCGA), that receives the video objects of segmented reference images, and outputs the corresponding m otion and layer information, using object and layer genotypes. Genetic operation strategies of reproduction, crossover, mutation, and domina nce [1] are applied recurrently in order to create successive generati ons of genomes with much better fitness, until convergence, or the max imum allowed number of generations is reached. For the increase of pre diction accuracy and convergence speed, lifetime fitness strategy [3] is used. Simulations with synthetic images have shown very encouraging results with the proposed video motion estimation technique, which co mpetes favorably with respect to conventional algorithms in accuracy, effectiveness, robustness, simplicity and speed.