An Overview of Run-length Encoding of Handwritten Word Images

Analysis of handwritten word images is closely tied to the method of representing the images. Diierent representations have their own sets of advantages and disadvantages. In this paper, we propose a novel method of encoding handwritten images using vertical runs that signiicantly simpliies the implementation of several image-processing tasks pertaining to handwriting recognition. We demonstrate the advantages of both horizontal and vertical run-length encoding schemes and compare them to other widely used representations like chain-code and bitmap. We illustrate ease of use of horizontal runs for correcting the slant angle, image smoothing, and base-line detection and vertical runs for correcting the skew angle and character segmentation. We believe this paper will serve as a useful tutorial in image representation schemes used in handwriting analysis and recognition.