Optimization of the Anisotropic Gaussian Kernel for Text Segmentation and Parameter Extraction

• Published in: IFIP International Federation for Information Processing/IFIP Vol. 323; pp. 140 - 152
• Main Author: Brodić, Darko
• Format: Book Chapter
• Language: English
• Published: Germany Springer Berlin / Heidelberg 2010; Springer Berlin Heidelberg
• Series: IFIP Advances in Information and Communication Technology
• Subjects: Document image processing; Image processing; Isotropic Gaussian kernel; OCR; Text parameters extraction; Text segmentation
• ISBN: 3642152392; 9783642152399
• ISSN: 1868-4238; 1861-2288; 1571-5736; 1861-2288
• DOI: 10.1007/978-3-642-15240-5_11

In this paper, extended approach to Gaussian kernel algorithm for text segmentation, reference text line and skew rate extractions is presented. It assumes creation of boundary growing area around text based on Gaussian kernel algorithm extended by anisotropic approach. Those boundary growing areas form control image with distinct objects that are prerequisite for text segmentation. After text segmentation, text parameters such as reference text line and skew rate are calculated based on numerical method. Algorithm quality is examined by experiments. Results are evaluated by RMS method. Obtained results are compared with isotropic Gaussian kernel method. All results are examined, analyzed and summarized. Furthermore, optimal parameter values are suggested leading to anisotropic kernel optimization.