00000ctm a22000004a 4500 UP-99796217603116283 Buklod 20230503092542.0 m |o d | ta 070713s2001 xx r |||| u|eng d (iLib)UPD-00000179724 DENGII eng LG 993.5 2001 C65 A88 Asuncion, Karen S. Optical character recognition for machine-printed characters Karen S. Asuncion ... [et al.]. 2001. 62 leaves ill. Thesis (B.S. Computer Science)--University of the Philippines, Diliman. In today's fast paced, on-line business environment, companies are faced with an explosion i the amount of data that must be processed and stored. Our group proposes to tackle this problem through optical character recognition. Our project's objective is to be able to "read" receipts and store them in a data file for archival and accouting purposes. This paper presents an efficient algorithm that recognizes a mixed document consisting of alphanumeric text, and graphic images. In the preprocessing step, an input document is skew-normalized, if necessary, by rotating it by an angle detected with the Hough transform. Then the graphic images are separated from the text parts by considering chain nodes of connected components. Each character is further separated using vertical and horizontal projections. In the recognition step, the text consisting of alphanumeric characters is recognized and this is where the Neural Network's task enters. A neural network is a generalized numerical tool which enables the correlation or linking of one set of data which shall be called the 'inputs' to another set called the 'outputs'. It is assumed that the inputs and outputs may be related in some way, although it is not necessary to know this relationship. Rather a known set of data, called the 'training set', containing both inputs and outputs for a number of different cases is used to teach the neural network to recognize any association which may exist. Therefore, the training set is said to comprise a number of 'patterns' each of which is a list of the inputs and outputs. Optical character recognition. Optical pattern recognition. Neural networks (Computer science). Gregorio, Genevieve G. Lucero, Krista C. Reyes, Joseph James C. Rivera, Artheo Day V. Thesis FI UP UPD DENG-II LG 993.5 2001 C65 A88 Thesis