Handwritten Hindi digits recognition of courtesy amounts in Arabian bank checks using neural networks

مقدم أطروحة جامعية

al-Khatib, Rammah Wajih Naji

مشرف أطروحة جامعية

Shilbayah, Nidal F.

أعضاء اللجنة

Hassan, Muhammad M. al-Hajj
al-Rawi, Kamal R.

الجامعة

جامعة الشرق الأوسط

الكلية

كلية تكنولوجيا المعلومات

القسم الأكاديمي

قسم علم الحاسوب

دولة الجامعة

الأردن

الدرجة العلمية

ماجستير

تاريخ الدرجة العلمية

2011

الملخص الإنجليزي

The problem of processing bank checks automatically started to emerge in banks as the need for a system that processes bank checks and recognizes handwritten fields increased.

In this thesis, a system is designed for recognizing the courtesy amount in Arabian bank checks.

This system is divided into two main parts; the first part was for the offline handwritten Hindi digits recognition system that recognizes separated Hindi digits using the Multilayer Perceptron Neural Networks with backpropagation training algorithm.

The samples were written by hand, 1000 of them for training and 600 for testing.

The samples were scanned and preprocessed before entering the Neural Network for recognition.

The system achieved a recognition rate of 74.4%.

The second part is for the courtesy amount recognition system.

The courtesy amount during this system passes through the segmentation and feature extraction stages.

The segmentation process divides the string of digits into separated digits and the feature extraction process extracts the digits so that they get ready to enter the recognition system in the first part.

Both stages performed properly and successfully.

التخصصات الرئيسية

تكنولوجيا المعلومات وعلم الحاسوب

عدد الصفحات

86

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-699363