I am a fan of using GDAL for image processing. It's very powerful and also very fast (useful when working with lots of images). The only downside is that it has a rather steep learning curve - it can take some time to become familiar with how it works.
As Pierre points out GDAL, particularly when combined with Python, especially NumPy and SciPy is a powerful combination, particularly if you have large volumes (100s-10000s of images) to process.
From what I gather most of the answers are related with land remote sensing correct? Because I only work with Satellite Oceanography (mostly ocean areas) and these are not the main software used. Just curiosity, do you process with any of these software you suggested above, ocean imagery?
As first choice for oceanography data I suggest the BEAM software. It was born to process ENVISAT onboard sensors' data, but now after 10 years of developments it supports quite all the sensors related to sea/water analysis and has a lot of tools for supporting calibration/validation analysis. Furthermore it supports also other sensors for land analysis and the support team is very fast in answering.
http://www.brockmann-consult.de/cms/web/beam/
Two more software I can suggest as complementary are:
OpenEV_FW which is a good tool to exploit the importing/exporting capabilities of GDAL
Hi, my suggestion would be the MultiSpec (https://engineering.purdue.edu/~biehl/MultiSpec/). For preliminary raster operations QGIS and TAS (http://www.uoguelph.ca/~hydrogeo/TAS/) is excellent. ILWIS is nice and rather simple but I had several crashes during my work.
I would have said SAGA GIS, Grass GIS and GDAL. I like GDAL because of the possible usage via Python where you can add any other third party library to it.