We present a novel genetic algorithm (GA) for video sequence segmentation. The novelty of the approach is that the mating rates such as crossover rate and mutation rate are not constant, but spatio-temporally varying. The variation of mating rates depends on the degree of activity of each chromosome in between the successive frames. The effectiveness of the proposed method will be extensively tested in the synthetic and natural video sequences and compared to several other GA-based segmentation method. The results show that the proposed approach is able to enhance the computational efficiency and the quality of the segmentation results than other methods.
展开▼
