03-26-2019 11:50 AM
We would like to pan-sharpen the imagery and orthorectify the pan-sharpened product. The imagery for pan-sharpening is multispectral (4 bands in a tif file format) and 1 panchromatic band (tif file format) processed to level 1B.
We have tried three methods to pansharpen the images. The first method we tried was ArcMap using the Gram-Schmidt algorithm. The other two methods were NNDiffuse and hyperspherical color space resolution merge in ERDAS IMAGINE. We then will orthorectify these using ERDAS IMAGINE. However, we have run into problems with these above methods:
Do you have any experience with these methods or have any suggestions on how we might fix these issues? Any comments and suggestions will be gratefully received.
03-27-2019 10:54 AM
NNDifuse pan sharpening method takes considerable time. We have addressed this by making use of GPU acceleration via OpenCL. This was done in ERDAS IMAGINE 2018 update 1.
If you have a Graphics Card that supports OpenCL, update your ERDAS IMAGINE to ERDAS IMAGINE 2018 update 1 or later and make use of the Open CL acceleration to spped up the processing.