Vectorization of raster images is done by converting pixel color information into simple geometric objects. The most common variant is looking over edge detection areas of the same or similar brightness or color, which are then expressed as a graphic primitives like lines, circles and curves.
A Raster graphics image is a rectangular grid of pixels, in which each pixel (or point) has an associated color value. Changing the size of the raster image mostly results in loss of apparent quality.
Vector graphics are not based on pixels but on primitives such as points, lines, curves which are represented by mathematical expressions. Without a loss in quality vector graphics are easily scale- and rotateable.