Handwritten word segmentation using Kaiser window

In the field of handwritten word recognition field, word segmentation into letters is an approach that could be used. Using this approach, word segmentation would be a complicated task, especially when dealing with a cursive handwritten word. A simple method in word segmentation called oversegmentation could be used. This paper discusses a simple method using Kaiser window. In general, the word segmentation process in this paper can be described as follow: Input - Preprocessing - Segmentation - Output. The input is an image of isolated handwritten word in binary format, while the output is images of letter segment. The main purpose of preprocessing is to correct slant and slope. This preprocessing is necessary since the segmentation method used is sensitive with slant and slope. The main purpose of segmentation is to divide a word into some letter segments. Based on a subjective test result, it was shown that the minimum parameters for the Kaiser window that can be used effectively for oversegmentation are 8 points in window´s length and 10 in beta value. As its window´s length is getting longer and its beta value is getting bigger, it can also be used effectively for oversegmentation. However, it must be noted that if the letter size is getting bigger, there will be more letter segments resulted.