Offline Character Recognition System Using Artificial Neural Network

摘要: Advancement in Artificial Intelligence has lead to the developments of various "smart" devices. The biggest challenge field image processing is recognize documents both printed and handwritten format. Character recognition one most widely used biometric traits for authentication person as well document. Optical Recognition (OCR) a type document analysis where scanned digital that contains either machine or script input into an OCR software engine translating it editable readable text A Neural network designed model way which brain performs particular task function interest. Each character comprised 30×20 pixels. We have applied feature extraction technique calculating feature. Features extracted from characters are directions pixels with respect their neighboring These inputs given back propagation neural hidden layer output layer. Back Network efficient errors were corrected through rectified neuron values transmitted by feed-forward method multiple layers.