When you download data from the providers such as United States Geological Survey (earthexplorer.usgs.gov), it will be in terms of individual spectral bands. Layer stacking is the process of creating a composite multispectral image from the individual spectral bands provided by the data provider. Layer stacking is the first and foremost process in image preparation. After layer stacking, generally resolution merge is done to improve the resolution of the image.

Subsetting and mosaicking are definitely preprocessing techniques and must be employed before classification. There are several reasons behind that. If you are trying to mosaic two images that are acquired at two different dates, then it a must to perform atmospheric corrections before mosaicking them. Similarly, there are so many other reasons (the number and types of corrections are too many to explain here). If you try to classify the data first and then try to mosaic them, then you would be trying to mosaic data that were classified under completely different input conditions. It is advisable to perform classification on the corrected then mosaic and subset image.

One of the reasons your ERDAS data not opening in ENVI may be due to compatibility. I suggest you to use EXPORT option in ERDAS IMAGINE to convert the data from .img to either .tiff or other Envi compatible formats. I suggest .tiff since it is a universally acceptable format across all RS/GIS platforms. You will find enough tutorials in youtube on these topics.

Hello,

yes,

some times the mosiacking need more color balancing which changes the DN values and affect classification.

layer stack is important when you have individual bands not multi spectral.

Dear Atrayee Roy ,

Mosaic related to georeferenced images, thus by classification this does not change and it is possible.