Reference text line identification using Gaussian kernel extended by morphological operations

In this paper, new approach for text segmentation, reference text line identification and skew rate estimation by Gaussian kernel is presented. As a result of basic algorithm, defined area exploited for text segmentation and text parameter extraction. To improve text segmentation process, basic method is extended by morphological operations. Furthermore, reference text line is estimated by least square method. Basic and extended algorithm is examined and evaluated under different text skew angles. Results are examined, analyzed and discussed. Proposed algorithm extension showed robustness for different types of skewness.