Abstract
Preprocessing is a very important part of cursive languages Optical Character Recognition (OCR) systems. Thus, baseline detection, which is one of the main parts of the preprocessing operation, plays a basic role on OCR systems; improvement on baseline detection could be absolutely useful for decreasing errors in recognition words. In this chapter, a metaheuristic- and mathematical-based algorithm is recommended, which has improved the baseline detection process in relation to the well-known baseline detection algorithms. The most important advantages of the proposed method are simplicity, high speed processing, and reliability. To test this novel solution, IFN/ENIT database, which is a well-known and attending database, is utilized. However, the proposed solution is reliable to any standard database of cursive language's OCR.
TopIntroduction
There isn’t an exact definition for the baseline concept of handwritten texts, however in our mind the baseline is a supposed line which in cursive languages (here Arabic), passes into the most words on a line. The baseline may serve for several usages such as: elimination or normalization skews (Al-Shatnawi & Omar, 2008; Pechwitz & Maegner, 2003), segmentation scripts into words or letters (Al-Shatnawi & Omar, 2008; Amin, 1998; Arica & Yarman-Vural, 2002) and to extract dependent features (Al-Shatnawi & Omar, 2008; El-Hajj et al., 2005). In Optical Character Recognition (OCR) systems, preprocessing is one of the most important parts of the system (Al-Rashaideh, 2006; Al-Shatnawi & Omar, 2008; Al-Shatnawi & Omar, 2009a; Farooq et al., 2005; Latfi et al., 2006) and the baseline detection, is a basic and necessary division of the preprocessing, too. Therefore, baseline detection is a very influential task for OCR systems, and if it does not work efficiently, it’s impossible to get an acceptable result. In other word it has a straight effect on accuracy and credibility of character recognition (Al-Shatnawi & Omar, 2008).
Generally, the aim of the current effort is to design an accurate, efficient and also simple baseline detection method for Arabic handwritten and typed texts, as by now there isn’t any perfect and reliable baseline detection technique yet.
However there are many difficulties and problems to design an accurate baseline detection method for cursive languages’ OCR systems. One of the most important related problems is that there are more letters with a non-geometric shape, so the places in handwritten texts are not totally evident toward the text baseline. Second or another problem is related to the sub-words. Each sub-word even in one word maybe have an own distinctive baseline. For example as it is shown in Figure 1, a given word with four sub-words could have four different baselines (AlKhateeb et al., 2011; Al-Shatnawi & Omar, 2008).
Figure 1. (a) A sample word with four sub-words; (b) the word with four separated baselines
Key Terms in this Chapter
Principal Component Analysis: A statistical procedure that uses orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of linearly uncorrelated variables called principal components. The number of principal components is less than or equal to the number of original variables. This transformation is defined in such a way that the first principal component has the largest possible variance (that is, accounts for as much of the variability in the data as possible), and each succeeding component in turn has the highest variance possible under the constraint that it is orthogonal to (i.e., uncorrelated with) the preceding components.
OCR System: A common method of digitizing printed texts so that they can be electronically edited, searched, stored more compactly, displayed on-line, and used in machine processes such as machine translation, text-to-speech, key data extraction and text mining. OCR is a field of research in pattern recognition, artificial intelligence and computer vision.
IFN/ENIT Database: The IFN/ENIT-database contains material for training and testing of Arabic handwriting recognition software. There are more than 2200 binary images of handwriting sample forms from 411 writers, about 26,000 binary word images have been isolated from the forms and saved individually for ease of access. A ground truth file for each word in the database has been compiled. This file contains information about the word such as the position of the words base line, and information on the individual used characters in the word.
Voronoi Diagram: A way of dividing space into a number of regions. A set of points (called seeds, sites, or generators) is specified beforehand and for each seed there will be a corresponding region consisting of all points closer to that seed than to any other. The regions are called Voronoi cells. It is dual to the Delaunay triangulation.
Baseline Detection: A process that allows you to find a baseline for the image using various methods.
Diacritics: A glyph added to a letter, or basic glyph. The term derives from the Greek d?a???t???? (diakritikós, “distinguishing”). Diacritic is primarily an adjective, though sometimes used as a noun, whereas diacritical is only ever an adjective. Some diacritical marks, such as the acute ( ´ ) and grave ( ` ) are often called accents. Diacritical marks may appear above or below a letter, or in some other position such as within the letter or between two letters.
Cursive Language: Languages in which the symbols of the language are written in a conjoined and/or flowing manner, like, Arabic, Persian, Cyrillic, etc.
Image Histogram: A type of histogram that acts as a graphical representation of the tonal distribution in a digital image. It plots the number of pixels for each tonal value. By looking at the histogram for a specific image a viewer will be able to judge the entire tonal distribution at a glance.