International Journal of Computer Applications (0975 – 8887) Volume 83 – No 10, December 2013 36 A Fuzzy Logic based Handwritten Numeral Recognition System Mahmood K Jasim Department of Mathematical & Physical Sciences College of Arts & Sciences University of Nizwa, Oman Anwar M Al-Saleh Department of Computer Science College of Sciences Al-Mustansiriyah University Iraq Alaa Aljanaby Department of Mathematical & Physical Sciences College of Arts & Sciences University of Nizwa, Oman ABSTRACT In this paper, a delayed treatment to handwritten numerals with fuzzy logic has been provided. The patterns which used in this system consisted 100 patterns of 10 numerals (0 to 9). They were taken from 10 different subjects and converted by the scanner to computer into 30×20 binary patterns. We used off-line system in take the patterns. The recognition rate is 94%. General Terms Image processing, Fuzzy logic Keywords Handwritten numeral recognition 1. INTRODUCTION Pattern recognition system is regarded as a system, whose input is the information of the pattern to be recognized, and output is a class to which the entered pattern belong [1, 2]. One of the important fields in pattern recognition is character recognition. It improves man-machine communication in addition to providing a solution for processing large amount of data automatically. The main objective of character recognition is the conversion of a graphical document into a textual one [2, 3]. Many systems have been proposed to recognize pattern. Some of these systems have been used the fuzzy logic. Most of the character recognition systems require preprocessing operations on the pattern [4, 5, 6, 7]. Preprocessing is an important step in pattern recognition systems in which fundamental features of pattern are extracted and/or enhanced, to classify and recognize unknown pattern [8]. Many tedious tasks can be made more efficient by automating the process of reading handwritten numerals. In such system an optical scanner converts each handwritten numeral to a digital image, and computer software classifies the image as one of the digits zero through nine. By reducing the need for human interaction, numeral-recognition systems can speed up jobs such as reading income tax returns, sorting inventory, and routing mail. Several steps are necessary to achieve this. A recognition system must first capture digital image of handwritten numerals. Before attempting to classify the numerals, some preprocessing image might be necessary. An algorithm must then classify each handwritten numeral as one of the ten decimal digits [8, 9]. Although a qualitative description of this process is straightforward, it cannot be easily reduced to a few simple mathematical rules. The difficulty results from the natural variations in human handwritten. A useful recognition system must be robust to alterations in size, shape, orientation, thickness, etc. Closed-form mathematical models tend to be inadequate for such a task because of the many possible representations of the same image. The paper presents an off-line system for the recognition of handwritten numerals with preprocessing steps uses fuzzy logic, and called Handwritten Numeral Recognition Using Fuzzy Logic (HRUFL). The rest of the paper is organized as follows. Section 2 provides the necessary details for constructing the proposed recognition system for handwritten numerals. The details of the HRUFL implementation have been presented in Section 3. Test result with the discussion is given in Section 4 while, Section 5 is the conclusion. 2. HANDWRITTEN NUMERAL RECOGINITION SYSTEM An Image of Handwritten Numeral (HN) converts to a binary image with global threshold (Level). The next step is the thinning processing to extract the skeleton of the HN, so the HN becomes ready to undergo isolation. The output of this part of HRUFL is the image of the character with a fixed image size; each character is assigned a recognition stage by fuzzy logic. All the experiments performed in this research deal with handwritten numerals from 0 to 9. The patterns which used in this system consisted 100 patterns of 10 numerals (0 to 9). They were taken from 10 different subjects and converted by the scanner to computer into 30×20 binary patterns. We used off-line system in take the patterns. The character set to be recognized appears in Figure 1 as a study case. There are ten valid characters-numeric characters zero through nine. Figure 2 shows the block diagram of the proposed system. Fig 1: Character Set to be Recognized. 2.1 Image Acquisition In numeral recognition systems, the first step is the acquisition of a digitized image of the text. Two methods are used in HRUFL in acquiring images. The first method is written the character by using paintbrush software tool, and converted to 2-D matrix. The second method uses a monochrome optical digital scanner.