基于GA—BP算法的多分辨率遥感影像融合技术  

A Study on A New Method of Multi-spatial-resolution Remote Sensing Image Fusion Based on GA—BP

作　　者：陈雯[1] 王远飞[1] 

机构地区：[1]华东师范大学地理信息科学教育部重点实验室,上海200062

出　　处：《遥感技术与应用》2007年第4期555-559,共5页Remote Sensing Technology and Application

基　　金：国家863高技术研究发展计划项目(2002AA134020)资助

摘　　要：由于Landsat-5唯一的热红外波段遥感影像TM 6的空间分辨率不高,使得其应用与研究程度远不及其它波段广泛。为此,运用GA—BP算法来提高TM 6遥感影像的空间分辨率,并进行仿真实验,结果表明:①GA—BP算法有效地避免了BP算法陷入局部最小点、收敛速度慢的问题,是一种快速、可靠的方法。它的快速算法对数据量巨大的遥感图像更具实用价值。②从提高TM 6遥感影像空间分辨率的仿真结果来看,无论计算效率还是遥感影像的融合效果,GA—BP算法都优于BP算法。③GA—BP算法既保留了TM 6遥感影像的基本灰度分布信息,同时也提高了其空间分辨率,可以有效地运用到提高遥感影像空间分辨率的过程中。Low spatial-resolution is a disadvantage of TM thematic remote sensing （TM6） image, which hampers its widespread application. So this paper put forward a new method based on the genetic algorithms and back-propagation algorithms （GA-BP） to improve the spatial resolution of TM6 image. This method combines the genetic algorithms and neural network, replacing the BP algorithm with the genetic algorithm. Tests showed that：①GA-BP algorithm can avoid the problems of BP algorithm which indicate calculation procedure is easily caught in local minimal point and the convergence is slow. The advantages of GA-BP are especially of more practical use for mass calculation of remote sensing images.②According to the simulation result, GA-BP algorithm had a distinct advantage over BP algorithm, no matter calculation efficiency or fusion effectiveness. ③GA-BP algorithm improves spatial resolution of TM6 image and the gray value can also be reserved, thus can be applied into the practice of improving the spatial resolution of TM6 image.

关 键 词：BP神经网络 GA—BP算法 图像融合 TM6 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]