Script-Independent Text Line Segmentation in Freestyle Handwritten Documents

Abstract

Text line segmentation in freestyle handwritten documents remains an open document analysis problem. Curvilinear text lines and small gaps between neighboring text lines present a challenge to algorithms developed for machine printed or hand-printed documents. In this paper, we propose a novel approach based on density estimation and a state-of-the-art image segmentation technique, the level set method. From an input document image, we estimate a probability map, where each element represents the probability that the underlying pixel belongs to a text line. The level set method is then exploited to determine the boundary of neighboring text lines by evolving an initial estimate. Unlike most connected component based methods [1,2], the proposed algorithm does not use any script-specific knowledge. Extensive quantitative experiments on freestyle handwritten documents with diverse scripts, such as Arabic, Chinese, Korean, and Hindi, demonstrate that our algorithm consistently outperforms previous methods [3,1,2]. Further experiments show the proposed algorithm is robust to scale change, rotation, and noise. Keywords: Handwritten Text Line Segmentation, Document Image Analysis, Density Estimation, Level Set MethodsThe support of this research by the Department of Defense under contract MDA-9040-2C-0406 is gratefully acknowledged. 1 Report Documentation PageForm Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number.

Segmentation of Handwritten Document Images into Text Lines

“…Consequently, the lobes (valleys) of the projection correspond to foreground (Li et al, 2008) with permission from the author.…”

“…The second step concerns the modification and use of well-known image segmentation methods in order to formulate the text lines. Li et al (2008) apply an anisotropic Gaussian filter to smooth the image and provide a greyscale "probabilistic" image that denotes the text line distribution. It is worth to mention that the horizontal dimension of the filter is greater than the vertical in order to advance the mainly horizontal text-line orientation.…”

Textline information extraction from grayscale camera-captured document images

“…In handwritten textline segmentation domain, Li et al [17] introduced the concept of straight textline enhancement of binary image using anisotropic Gaussian smoothing. This textline enhancement approach can not be applied on camera-captured document images which contain nonstraight textlines with high degrees of multi-oriented curl and variable font sizes.…”

Detection of Text Lines of Handwritten Arabic Manuscripts using Markov Decision Processes

“…The Arabic script is naturally cursive, which makes the extraction of lines from handwritten Arabic document a real challenge [2]. Among the challenges, we have:…”