Investigation of Resampling Effects on Irs-1d Pan Data

Image resampling is frequently employed in geometric correction as a part of preprocessing of a remotely sensed image. The resampled images will have a noticeable change in image quality, which may have an adverse impact on the accuracy of subsequent image analysis for information extraction. Information of edges is very important, especially in high spatial resolution imagery. The present study has investigated the effects of resampling on edge information for Indian remote sensing (IRS)-1D panchromatic (PAN) imagery with 5.8 m spatial resolution. Raw PAN images have been resampled during shift resampling and geometric rectification processes by using bilinear, cubic B-Spline and cubic convolution resampling methods. The cubic convolution was implemented with three different parameter values of a as -1.0, -0.75 and -0.5. The resampled images were then used to identify edges. It has been observed from the statistical properties and by using edge detected images that the cubic convolution method of resampling with a as -0.75 value is the most suitable resampling approach for IRS-1D PAN imagery, especially for detecting edges.