Abstract :
A novel genetic algorithm for video sequence segmentation is presented. The novelty of the approach is that the mating rates such as crossover rate and mutation rate are not constant, but spatio-temporally varying. Experimental results demonstrate the effectiveness of the proposed method.