Hi Nadiok, Did you try ENVI 5 and the image-to-image registration workflow from the Toolbox. What are the exact steps you following and what parameters are you setting?

Oh, ye, i already tried with image-to-image registration, but i really do not understand, how can i interpret result. But i was not quite sure, that this is in general - right method for my purpose. You mentioned PARAMETER, but i have no parameter, or i dont know what does it mean. because i have 5 Rapideye-scenes with same map information and extension and they are time-series -mages(5 meter spatial resolution, UTM WGS 84, 32 Zone). My task in my thesis is ( as first importang step), that i have to match these 5 images with each other in order to combine them in classification process to get one final file. Because i know from my professor, that these images are not really matched. In order to avoid later relocation of crop fields in final classified file- i have to match them first of all. I already took a look on SPEARS- tool for coregistration...is it also good idea?Thank you sooo much!

Maybe you should contact Exelis VIS technical support with your question/problem. Let them know all the steps you are using and what parameters you are using. Parameters are items that you enter in the dialogs for any steps you follow. Don't forget to include your license number. If you don't know how to find that the following article will help: http://www.exelisvis.com/.../ArticleId/4643.aspx

Unfortunately i cannot ask technical support there, because they said to me, it is not possible, i have only student edition of ENVI 5. Can you help me to give any tip if for my purpose these methods good are: SPEAR TOOL- image to map registration (coregistration) and also MAP- REGISTRATION-IMAGE TO IMAGE REGISTRATION?? Because this is my deal, i am lookinf in ENVI 5 student edition for right methods for matching my multitemporal data. Because my scenes are already georeferenced, but not matched with each other. Maybe i can do follow thing: i take one image of 5 images and use it as BASE IMAGE for ALL OTHER IMAGES (from different time-point) and use IMAGE TO IMAGE /IMAGE TO MAP REGISTRATION? i actually can match only pair of images as i see in ENVI. Thanks a lot for your help!!!!

It sounds like any of the image-to-image registration tools would be appropriate for your goals. If you happen to have good map coordinates for a lot of the features in your scenes (say, from a vector layer, or a georeferenced image, or a GPS), then image-to-map registration could also be appropriate. If you are having trouble getting good results from registration tools, can you provide more information about exactly what you are doing (including which tools you are using), and exactly what seems wrong with your results? - Peg

Hello Dear Peg, everything i have- 4 RapidEye scenes for later land use classification. My first important step is to match these multitemporal scenes (these scenes are from 2011: May, Juni, August..etc..), because my superviser said, they are not really matched, but they are already georeferenced in general, so that they only should be matched with each other.Yes, the image-to-image registration seems really to be appropriate for this goal. But what i do not know is: 1) i have these scenes as OVR-datafile and also metadata.xml-datafile. Which of them should i use for registration? (rapideye has 5 bands and 5 meter spatial resolution, UTM, WGS 84, 32 Zone) 2) Is it good idea if i take one of these scenes as base-image and use it for all other scenes as BASE-INPUT-IMAGE?i guess, i choose GCP-s for this base image and save it and use always for every pair of images (base and wharp)? 3) With bad results i mean maybe not really bad results, but i am not quite sure i can distinguish different results..with dynamic overlay they seem to be all very similar...should i only use this dynamic overlay for registration accuracy? I have heard that for later supervised classification for land use map from these scenes- the best choice is the nearest-neighbor-resampling-method? I would very appreciate your help in these 3 points!!! Thanks in advance!

To open a RapidEye image in ENVI, you will want to choose the RapidEye reader under File > Open As (if you are in ENVI 5). That reader expects you to specify the *.XML file associated with the dataset. Once you have read all of your RapidEye images into ENVI, then it does make sense to choose one of your images to be the base image, and register all of the other images to that one. You will need to choose different GCPs for each image that you register to this base image, though. Because the GCP files contain not only the coordinates for the base image, which could be the same for each registration, but also the coordinates for the warp image, which will be different for each image that you wish to register to the base. Well, if you use the dynamic overlay, and the two images (base and warped) look very close, then I think that means that your registration is good. Yes, most people believe that when you register imagery, if you plan to do any subsequent processing that is based on the pixel values (like classifications) then you would want to choose a nearest neighbor resampling. It will leave your images looking less spatially smooth, but it will preserve the original pixel values, which are what is actually measured by the sensor. - Peg Exelis VIS Tech Support