Feature Extraction and Image Recognition of Cursive Handwritten English Words Using Neural Network and IAM Off‐Line Database

Summary

The Neural Network implementation of cursive handwritten English words, in this paper, aims to convert an individual handwritten word to digital format. The Neural Network consists of layers of Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and Connectionist Temporal Classification (CTC), which can be trained on the CPU. The system proposed does not segment the input, but rather the layers extract relevant features from the scanned images fed as input. Compared with previous systems for handwritten text recognition, the given architecture is end‐to‐end trainable and does not require different components to be trained separately. It naturally handles sequences in random lengths, involving no horizontal scale normalization or character segmentation. The model is smaller yet effective, thus, more practical for application in real‐world scenarios.