遥感图像亚像元定位元胞自动机算法改进研究  

Modification of Cellular Automata(CA) Used in Sub-pixel Mapping

作　　者：李晓俊[1] 吕恒[1] 李云梅[1] 王瑜[1] 张杰[1] 潘洪洲[1] 

机构地区：[1]南京师范大学虚拟地理环境教育部重点实验室,江苏南京210023

出　　处：《遥感技术与应用》2015年第6期1206-1214,共9页Remote Sensing Technology and Application

基　　金：国家自然科学基金项目"微囊藻生物光学特性及其遥感定量识别研究"(41171269)

摘　　要：元胞自动机模型(CA)常用于图像亚像元定位,它将混合像元剖分为亚像元,再利用端元组分的丰度及像元空间分布的特点,将亚像元赋予不同端元组分得到各端元组分的空间分布情况。但其存在迭代次数过多、进化过程"不充分"以及结束迭代的条件不明确等缺点。针对上述问题,对亚像元定位CA算法进行了3方面的改进。结果表明:最终的亚像元定位结果调整后Kappa系数虽然平均只提高了4.1%与3.7%,但迭代次数大幅减少,均在5次以内得到最优结果;运行时间大幅下降,仅占原方法的25.8%,使得CA的运行效率得到了显著提高。此外明确了CA的终止条件,使得面向亚像元定位的CA算法更具实际应用价值。Cellular Automata（CA）was frequently used to Sub-pixel mapping,which is a technique designed to obtain the spatial distribution of classes with information contained in mixed pixels.However,it has several drawbacks,such as large number of iterations,evolution ＂insufficiency＂and equivocation of termination condition,which restricted the further application in remote sensing.In this study CA was improved in three aspects to overcome problems that mentioned above,which included the modification of initialization,the modification of CA progress and a clear termination condition.Two images were employed to test running efficiency of improved CA.Although the adjusted Kappa coefficients of final Sub-pixel mapping results increased only by 4.1% and 3.7%,number of Iterations reduced and the CA running time accounted for only 25.8% of the original.The result showed that the improved CA can significantly improve the running efficiency.In addition,the clear termination condition made it more practical application value.

关 键 词：遥感 亚像元定位 元胞自动机 尺度 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]