Writer Identification from Gray Level Distribution

When identifying a writer from a handwritten text, most often, either some characteristic patterns or some shape parameters are extracted. They are assumed to be specific of the writer. Here, we are to explore a different approach, we consider the distribution of the pixel gray levels within the line. It is linked to pressure and writing speed when text is realized. In the line, the direction that is perpendicular to the writing way of drawing is privileged. The curve associated with the gray levels in a stroke section is characterized by use of 4 shape parameters. More over the regular sections are selected and are grouped in section lots. The distributions of the sections and of the section lots are quantified. Thus 22 parameters are extracted. Three different classifiers are used with and without genetic selection of the most significant parameters for the classifier. Then the classifiers are combined and the results show the gray level distribution within the writing is characterizing the writer in a significant way. keywords : writer identification, gray-level image, gray level distribution, stroke section.