Factors affecting accuracy of binarization for document images

Scanned and camera captured document images may contain 256 grey levels, which needs large computations for text extractions. Binarization lowers these computations by converting 256 grey levels into two levels (0/1). General image binarization function can be expressed as given in Equation

g(x,y) = Op f(x, y)

Where f(x, y) is the input image, g(x, y) is the processed image and Op is an operation applied on f, defined over some neighbourhood of (x, y). As our technical era is growing, Image capturing, storing and extraction method are changing very rapidly. These changes affect the accuracy of existing Binarization techniques to great extent. That's why Binarization is prominent area of research from last two decades. This paper discusses about various factors which affects the accuracy of binarization obtained by observing various document images and noises which may degrade the quality of a document image.