Definitions

API: This stands for Application Programming Interface, libraries of functions that a programmer uses to create applications.
CSS Pixel: A pixel that maps to one or more device pixels.
em: The em is not the height/width of the letter ‘M’, or even ‘m’. The em is a property of TrueType fonts and one em is the font size. The name is a historical carryover from print typography.
Image dpi: Some graphic files have numeric fields for storing image horizontal and vertical dpi values. These values are there for the purpose of determining the size of the printed image. Programs that display an image on the screen ignore dpi values stored in a graphics file so statements like “Save images for the web at 72 dpi” are nonsense. Note that the popular GIF format doesn’t even have a field for storing a dpi value. A dpi value stored in a graphics file tells one nothing about the quality of an image unless you know that the image was obtained from a scanner operating at that many samples per physical inch. Suppose I give you an image that my graphics program says has a ‘resolution’ of 96 dpi. Being sneaky, I could then give you the identical image but just change the stored dpi value to 300 dpi. Would you really conclude that this second image has better resolution, i.e. more detail, then the first image? I hope not.
Scanner dpi: A desktop scanner driver knows the physical size in inches of its work surface and, hence, software can get the physical size of a document or photograph (provided it can detect the edges). The dots are pixels and the inches are physical inches. The dpi value is stored in the image file, depending on the file format, and is solely for defining print dimensions.
Screen dpi: Screen dpi has one purpose, and one purpose only, and that is to transform screen inches into screen pixels. It has nothing to do with physical inch measurements or screen images. The dots in screen dots per inch are pixels, but the inches are logical inches and not physical inches. It would make more sense if we called it ‘Screen ppli’ (pixels per logical inch), but who wants to be sensible. Whenever you specify output to a screen in inches then you are using logical inches and not physical inches. If you are using the Windows default setting of 96 dpi ( 96 pixels per logical inch) and specify it to be one inch wide then it will be 96 pixels wide.
Screen ppi: Screen pixels per inch is the ratio of screen physical pixels to screen physical dimensions. If your maximum video resolution is 1280 x 1024 then you have 1280 physical pixels horizontally and 1024 physical pixels vertically. That’s easy for getting the number of physical pixels, but how do you get the screen’s physical dimensions? The answer is ’with great difficulty’. The only way to get true physical screen dimensions is to measure the screen with a ruler! Having said that, one can actually get screen physical dimensions via the API but not accurately.
Square Pixels: This is a term one sees from time to time and has nothing to do with the physical shape, actual or perceived, of a pixel. If Horizontal Screen ppi = Vertical Screen ppi then you have ‘Square Pixels’, if not then you have ‘Non-Square Pixels’. If you are talking about Square Pixels in relation to a TV screen then you are on your own, otherwise forget about it.
Video Resolution (a.k.a Screen Resolution): This is the how many pixels the video card displays in the horizontal and vertical directions e.g. 800 × 600, 1024 × 768.