Preprocessing Framework for Document Image Analysis

Preprocessing is the first step used in all the document image analysis algorithms. A well organized preprocessing could lead to better results of the analysis. This paper proposes a framework for preprocessing of document image for analysis. The frame work uses four steps such as color image to grayscale conversion, enhancement of grayscale image, binarizing the grayscale image and finally removal of clutter-noise. Horizontal and vertical projections are used to detect possible locations of clutter noise in this work. Then foreground pixels are replaced by background colored pixels based on the run length. The frame work provided better results for test images.