This paper describes a new, very efficient video object motionestimation strategy, that considers video object translation,translation of rotation center, and planar multilayering. It is based ona Multipopulation Modified Coevolutionary Genetic Algorithm (MMCGA),that receives the video objects of a segmented sequence of video images,and outputs the corresponding motion and layer information usingappropriately represented object and layer genotypes. The algorithmperforms a random search for locating the global optimal solution in thesearching space. This operation involves an evolutionary process whereinpopulations of predicted solutions evolve over a period of generations.From the possible solutions, a video test frame is created, and thefitness of the test frames is evaluated by comparing with the referenceframe. Next, solutions are selected for reproduction based on theirfitness values. Good solutions are selected for reproduction while badones are eliminated. The selected solutions undergo recombination underthe genetic operations of modified crossover and dynamic mutation. Thislast operation increases the mutation rate and reduces the mutationrange from one population generation to the next, maximizing theperformance of the MMCGA. For the increase in predicted solutionaccuracy, and convergence speed, lifetime strategies are used.Preliminary simulations with synthetic video images have shown veryencouraging results with the proposed video motion estimation technique,which competes favorably with respect to the conventional motionestimation algorithms in accuracy, robustness, simplicity andspeed
展开▼