 The image above illustrates the process of optical character recognition on the basis of an 8 dot pixel matrix featuring the phonographic alphabetic sensorif letter P. At the lowest level of analysis in a feature detector model, geometric features, for example horizontal lines, edge location, pixel position, color saturation, etc. are analyzed. At the level of sub-graphemic orthographical features, loops and concavities will be detected. Here a P can be differentiated from R. The evidence collected at this point is dropped into a feature buffer where it remains for a short interval until further information comes in to combine the features into characters.