When searching for images, there are usually two big challenges:
Finding the image:
To find an image, most search engines rely on file names, image tags or accompanying text to tell what the image is. We're really not at the point yet where a search engine can look at a picture and tell that it's an elephant -- it needs human entered text to it that. This means relying on whoever put the image on the database or Internet to accurately describe the image. Images with vague titles such as "untitled" or without any text are images you're likely to never find.
Knowing if you can use the image:
Another problem is knowing whether you are allowed to use the image for your own purpose or not. There are a lot of guidelines about images and the rules are not always clear. For more info, check out the copyright section of this guide. Also whenever you are in doubt, check with the image owner (if you can find out who that is).
Credit: D. Medrano
Databases tend to be more reliable than the Internet in terms of image results. This is because there are a group of professionals behind the scenes making sure every image is properly titled and tagged. Another benefit is that most databases will give you the usage rights within the image record (in other words, will let you know if you can legally use the image). The drawback to databases is that they tend to include much fewer results than Internet searches.
If you're looking for quantity, the Internet is the way to go. There are thousands to millions of results for any given topic. The problem is that you must rely on who created the web page and uploaded the image to accurately describe it. It is very possible to come across an image that has absolutely nothing to do with your search term. In some cases, the image may be pornographic -- even with filters on. Other problems with the Internet include unclear ownership and usage rights and unreliable image quality.