PENGEMBANGAN APLIKASI TEXT RECOGNITION DENGAN KLASIFIKASI NEURAL NETWORK PADA HURUF HIJAIYAH GUNDUL

Abstract

Technological character recognition or OCR (Optical Character Recognition) have had fairly rapid growth and many have been used. Although the text recognition programs are widely available, but the majority of these programs can only recognize the Latin alphabet. How to recognize Arabic characters are converted to text alphabet is still very rare. Though the existence of Arabic text recognition has many advantages, for example, for people who do not recognize the Arabic script, it can be used as an alternative way to learn. This text recognition application using Artificial Neural Network learning algorithms (ANN) which is backpropagation algorithm an effective character classification which described in this study. In this study, the application will be developed to recognize the character recognition of Arabic characters Hijaiyah bare particular letters and Arabic numerals. Algorithms presented in the program as a tangible example of OCR (Optical Character Recognition) will be implemented in the Java programming language (Java Programming Language). With a certain level of accuracy in operation, can be obtained results (output) character appropriate to the class learning formed through the "OCR Train". Once fully developed, the application will then be tested by establishing Hijaiyah characters and Arabic numerals which are stored as image / image. Then open the file with the appropriate type of artificial neural expected to recognize the text in images of text characters / images are tested.
Keywords: Optical Character Recognition, Text Recognition, Java Language Programming, Artificial Neural Network, Backpropagation, Arab.