摘要
为解决林区缺乏大比例尺地形图而难以进行高空间分辨率遥感图像几何精校正问题,采用基于广域差分技术的星基差分GPS精确测定地面控制点(GCP),以多项式变换进行QuickBird pan(0.61 m)图像几何精校正.结果表明,全部GCP的点位误差均小于1个像元,总的点位误差为0.411 9.在进行图像灰度重采样中,最邻近点法、双线性内插法和三次褶积法的效果均无明显差异;就信息量和灰度值变动情况而言,以三次褶积法略好;就图像灰度保真而言,以最邻近点法略好.
To resolve the problem of geometric correction of high-resolution remote sensing images due to the lack of large scale relief maps in mountainous areas,a wide area differential GPS system(WADGPS) was used to survey the ground control points(GCPs) and to correct the QuickBird HRGs panimage(0.61 m) through the model of polynomial.The results indicated that the RMS of all the GCPs were smaller than one pixel and the total RMS is 0.411 9.In the resampling of images,no marked difference existed among the methods of nearest neighbor,bilinear interpolation and cubic convolution.The cubic convolution was a little superior to the other two methods in information content and the standard deviation of pixel gray values,and the nearest neighbor was the best method in preserving the image gray values.
出处
《北京林业大学学报》
CAS
CSCD
北大核心
2005年第S2期48-51,共4页
Journal of Beijing Forestry University
基金
"863"国家高新技术项目(2003AA245030)
国家自然科学基金项目(90302014
30371159)
北京市自然科学基金重点项目(4041002)
