节点文献
差分进化在图像特征选择中的应用研究
Research on the Application of Differential Evolution for Image Feature Selection
【摘要】 图像特征选择是图像检索至关重要的一个环节,本文将差分进化计算应用于图像特征选择中。针对差分进化计算存在过早收敛的问题对其进行改进,给出了改进算法的交叉、变异、选择操作过程,同时将两种算法应用于图像检索中。利用Corel标准图片库测试算法的性能,结果表明改进算法的运行时间与原算法相当,但其查全率与查准率明显优于原算法。
【Abstract】 Image feature selection is a crucial link for image retrieval,and the differential evolution is applied to image feature selection。An improved algorithm is gived because differential evolution has the problem of premature convergence.Also crossover,mutation,selection process is gived detaily.Two kinds of algorithms has been used in image retrieval.The performance of both algorithm is tested using he Corel standard picture library.The results show that the improved algorithm almost has the same running time compared to the original algorithm,but it has better retrieval accuracy.
【Key words】 differential evolution; feature selection; crossover and mutation; retrieval accuracy;
- 【文献出处】 科技通报 ,Bulletin of Science and Technology , 编辑部邮箱 ,2013年08期
- 【分类号】TP391.41
- 【被引频次】3
- 【下载频次】116