Character Recognition for Cursive English Handwriting to Recognize Medicine Name from Doctor's Prescription

This paper aims to represent the work related to character recognition for cursive English handwriting recognition. Recognition of cursive characters is very challenging because the characters are connected to each other. In the proposed architecture, horizontal projection method is used for text-line segmentation and vertical projection histogram method is used for word segmentation. Convex hull algorithm is used for feature extraction and SVM is used for classification. Proposed work is specifically for medical domain to recognize the medicine name from doctor's prescription. The segmentation experiments are carried out on different doctor's prescription samples and achieved 95% accuracy for text-line segmentation and 92% accuracy for word segmentation. The recognition accuracy of proposed system is 85%.