I'd recommend you read some of the tutorials here (even if it looks old-school), as it gives visual examples of each algorithm's output: http://homepages.inf.ed

I'd recommend you read some of the tutorials here (even if it looks old-school), as it gives visual examples of each algorithm's output: http://homepages.inf.ed.ac.uk/rbf/HIPR2/wksheets.htm http://homepages.inf.ed.ac.uk/rbf/HIPR2/wksheets.htm You could also try to read the documentation/source code of some open source image analysis software, and prototype your application around them: http://opencv.org/ http://opencv.org/ http://hugin.sourceforge.net/ http://hugin.sourceforge.net/ Once your prototype works as expected, you can learn more about the algorithms you used, and implement it yourself in your language of choice. BTW when reading the title of the article I expected something about Debian's package manager, aptitude :) Edit: to answer to the more general question in the post: if a problem is interesting enough I would invest some time in learning how to solve it. If it turns out that the domain (image analysis in your case) is not as interesting as I thought, or would require a significant amount of time to learn properly then I'd probably avoid that domain for a while. But I wouldn't give up on it just because I didn't learn about it in school.