摘要
影像质量评价对遥感影像处理起着非常重要的作用。本文在论述当前常用全参照遥感影像质量评价指标的基础上,选用IKONOS,TM,SPOT,QuickBird四幅光学影像作为实验原数据,首先对各影像添加不同程度的高斯白噪声和降质重采样处理,然后以原数据作为全参照进行统计对比分析,验证信息熵、清晰度、均方误差、峰值信噪比、结构相似度5个常用影像质量评价指标的可靠性。通过对实验结果的分析与总结,得出了一些有益结论,可为光学遥感影像的选取与处理提供参考。
Image quality evaluation is very important to the processing of remote sensing images. The IKONOS image, TM image, SPOT image and Quick Bird image were selected as test images on the basis that the commonly used quality evaluation methods of remote sensing image by full reference were described in this paper, and then the four images were added by different levels of Gaussian White noise and resampled to degrade the resolution. The reliability of commonly used quality evaluation indexes such as Entropy, Definition, MSE, PSNR and SSIM were verified by the analysis and statistics of original data which was used as full reference. Some useful conclusions were summarized through the analysis of the test results, which could provide effective reference for the selection and processing of optical remote sensing images.
出处
《测绘与空间地理信息》
2014年第9期35-38,41,共5页
Geomatics & Spatial Information Technology
基金
国家科技计划(973计划)项目(2012CB719901)
国家测绘地理信息局重点实验室开放基金(DM2013SC03
GC201406
K2014B008)资助
关键词
高分光学遥感影像
全参照
客观质量评价
high resolution optical remote sensing image
full- reference
objective quality evaluation