Video segmentation and summarization based on Genetic Algorithm

This paper presents a Binary Genetic Algorithms (BGA) based video summarization system. The similarity functions are first defined to evaluate segmentation, which are extremely expensive to be optimized with traditional methods. Then the system employs binary crossover and mutation operators to get the meaningful summary in a video search space. In order to test performance of the BGA method, we first compare the BGA method with Decimal Genetic Algorithms (DGA) method. The obtained results show that it is more quickly to find the best results for BGA than DGA. Second, the BGA method and the uniform approach have been compared. Experimental results show that the BGA method can capture more information than the uniform method and reduce redundancy.