遥感影象融合效果的客观分析与评价

本文综合利用三类统计参数来进行遥感影象融合效果的分析与评价：第一类反映亮度信息，如均值；第二类反映空间细节信息，如方差、信息熵、交叉熵和清晰度，其中给出了一种更切合实际的交叉熵定义；第三类反映光谱信息，如扭曲程度、偏差指数与相关系数。以小波变换融合方法为例，针对TM多光谱影象与SPOT全色影象融合的具体情况，，对不同分解阶数的融合影象进行分析和评价。评价结果与理论分析结果和目视效果相吻合，说明这些参数可以作为遥感影象融合的客观效果评价准则。

Objective Analysis and Evaluation of Remote Sensing Image Fusion Effect

In this paper,three types of statistical parameters are synthetically used in the analysis and evaluation of remote sensing image fusion effect. The first type of statistical parameter reflects the intensity information, such as mean; the second one reflects the spatial detail information, such as variance , entropy, cross entropy, and definition , where a more practicable concept of cross entropy is presented; the third one reflects the spectral information , such as distortion degree , distortion index , and correlation coefficient. These statistical parameters are applied to evaluate fused images correspondingly with different decomposition levels in the fusion of a Landsat-TM321 (RGB) multi-spectral image and a SPOT panchromatic image with a wavelet transform fusion algorithm. The evaluation result coincides with the visual effect and the theoretical result,that is to say,these statistical parameters are suitable for the evaluation of remote sensing image fusion effect.