摘要
以高分辨率的QUICKBIRD卫星数据为研究对象 ,经反差增强和比值运算 ,使处理后的图像分辨率高且接近真彩色 ,具有较好的识别效果 ;在图像上均匀布设 1 3个地面控制点 (GCP) ,另选取 3个国家三级控制点参与静态GPS外业实测及内业解算 ,平差后各控制点坐标误差在 1cm左右。利用遥感图像处理软件ERMapper6 .2为平台 ,进行二次多项式几何精校正 ,校正时各控制点的RMS值控制在 1个像元以内 ;用 1∶5 0 0地形图对纠正后的图像进行精度检验 ,其点位误差在 1m左右 ,说明纠正后的图像具有较高的精度。本文还对纠正后的遥感图像进行了应用研究 ,结果表明 :各地类的判读精度均很高 ,其综合判对率为 95 .7% ;该图像能将针叶树种和阔叶树种分开 ,树种的识别仍有难度。通过纠正后的图像与已完成的研究区内土地利用现状图进行匹配与精度分析 ,发现土地利用现状图有的区域精度较高 。
This paper focused on the Quickbird data processing and its application.The processed image is up to real color, high resolusion and easy to be identified after contrast enhancement and ratio processing. 13 GCPs placed in the raw image evenly were measured by GPS in the field; another 3 points have also been selected as known points.The error of each measured point is about 1 centimeter, this showed that the GCPs are very accurate. The image was precisely corrected by ER Mapper 6.2, the RMSE were controlled to no more than 1 pixel; 1∶500 topographic map was used to check this corrected image to different appeiccdions, the displacement is about 1 meter, so the corrected image is very precisely. This paper also applied the corrected image to different applications, the result showed: the interpretation ratios of all sorts of land types are very high, but it is still difficult to identify the species of trees on this image; the composite interpretation ratio is 95.7%. The corrected image have been matched to the landuse map and the precision has also been analyzed, the result showed that some present landuse division is not accurate enough.
出处
《遥感信息》
CSCD
2004年第2期20-23,i002,共5页
Remote Sensing Information
基金
湖南省杰出中青年专家科技专项计划资助项目 ( 0 2JJYB0 0 5 )
湖南省教育厅科研资助项目 ( 0 1C2 5 9)
中南林学院青年基金重点资助项目 ( 0 418)
关键词
遥感图像
反差增强
假彩色
判读标志
土地利用
remote sensing image
contrast enhancement
psudocolor
interpretation sign
landuse
