Topic: about reference image and orthorectification

today I communicate with Sebastien, he think that our exchange maybe help  other users, so I post it on the forum.

Firstly, thank you for your kindly providing the sample data! I'm a new user for COSI-Corr and I learn how to use it through cosi-corr-guide.pdf, when I made tie points between my master and slave image. There is something wrong. perpaps I'm  confused with what you call the "reference image", or sometimes the "master image".  There is no "reference image"(ortho-image) in the sample data package?
In the sample data provided, you have to build your own master image using one of the two images provided. You can collect control points between the image from 1998 and a shaded version of the DEM to make sure of the image overall georeferencing, then produce the ortho-image, and use this first ortho-image as the master image for the second raw image.
Thank you for your reply! In light of your suggestion, I have completed the typical processing chain by using the sample data. But I still have some questions:
1. When I do not have any reference ortho-image, I need to create a shading version of the DEM; does this method work well when the resolution of my DEM is very low(Say, SRTM, 90m)?
2. In the user_guide, I found that the level 1A data is required for SPOT, then we start the processing chain from SECTION5.1(in the user_guide). If I have two ortho-image(GeoTIFF format,one pre-earthque, the other post-earthquake), may I omit the orthorectification step and calculate their correlation & displacement from SECTION9 ?
Using a shaded version of the DEM as the first reference image works only if the DEM resolution is not too distant from the image resolution.  It usually works poorly with SRTM 90m. Tie points / GCP can nevertheless be selected, but if the DEM resolution is too coarse compared to the image one, the GCP optimization is not likely to succeed.

If you have two ortho-images, you can directly try the correlation but you have to make sure that they are both lying on the exact same grid or you'll get an error at the beginning of the correlation. We can discuss about this further on the Forum if you run into this pb.

We do start from the 1A images because accurate geometrical processing is very important. When you are usually provided with ortho-images, they are typically not well registered and the correlation may show strong artifacts due to topography and sensor mis-corrections.


Re: about reference image and orthorectification

I don't understand why ,if I orthorectified the first spot image with Envi tool and then i use this like master image to orthorectified the second spot image, i will have an error of correlation due to Overlap...also if i had choose ADJUST... in map grid?