传统的基于进化聚类方法在处理变化检测时耗时过长，在搜索最优聚类中心过程中容易陷入局部最优，对于SAR图像的变化检测存在边缘定位不够准确的缺点，提出了基于量子免疫克隆聚类的SAR图像变化检测方法.把图像的灰度值作为输入信息，通过量子比特定义聚类中心，通过量子免疫克隆算法来搜索最优聚类中心，从而得到更佳的全局阈值，最后根据阈值得到变化检测结果.实验结果表明,与K&I阈值法相比，可以得到更佳的全局阈值；与遗传聚类算法相比，可以快速、有效地搜索到更优聚类中心，准确定位边缘，提高变化检测精度.
As the conventional evolutionary clustering optimization methods are often time-consuming and easy to trap in local optimal value in dealing with the problem of change detection. Furthermore, it can not detect the edge accurately for SAR images. We proposed a method for change detection in SAR images based on the clustering analysis. The proposed method takes gray-levels as an input, uses the quantum bit to define the clustering center, searches the optimal cluster center using the quantum-inspired immune clonal algorithm, and gets the global threshold. Finally, the change-detection map is produced. Compared with K&I threshold, it can achieve a better value. Compared with Genetic Algorithm Based Clustering (GAC), the proposed method can search a much better clustering center quickly and effectively. Besides, it can detect the accurate edge and improve the change detection accuracy.
李阳阳,吴娜娜,焦李成,尚荣华,刘若辰.基于量子免疫克隆聚类的SAR图像变化检测[J].红外与毫米波学报,2011,30(4):372~376]. LI Yang-Yang, WU Na-Na, JIAO Li-Cheng, SHANG Rong-Hua, LIU Ruo-Chen. Change detection for SAR images based on quantum-inspired immune clonal clustering algorithm[J]. J. Infrared Millim. Waves,2011,30(4):372~376.]
复制
