Application of Neural Network In Handwriting Recognition
Handwriting recognition can be divided into two categories, namely on-line and off-line handwriting recognition. On-line recognition involves live transformation of character written by a user on a tablet or a smart phone. In contrast, off-line recognition is more challenging, which requires automatic conversion of scanned image or photos into a computerreadable text format. Motivated by the interesting application of off-line recognition technology, for instance the USPS address recognition system, and the Chase QuickDeposit system, this project will mainly focus on discovering algorithms that allow accurate, fast, and efficient character recognition process. The report will cover data acquisition, image processing, feature extraction, model training, results analysis, and future works.
Image Processing Flow
Research Paper Link: Download Paper