Recognition of Tamil handwritten text from a digital writing pad using MWDCNN

Text recognition from the Tamil Digital handwritten alphabet is challenging for researchers. Recognition systems face numerous challenges, such as more bends, curves, and rings. More errors occur in recognition of Tamil words written in digital pen and pad due to angles and curves in the Tamil alphabet, which need to be accurately converted. This paper proposes a Multi-Resolution Wavelet Deep Convolution Network (MWDCNN) for Tamil handwritten text recognition from a digital writing pad. The MWDCNN enhances low-level and high-level character features for depth information from the images. The proposed algorithm is compared with traditional algorithms such as CDBN, CapsNet, DBN, CRNN, and CNN digital writing pad-based handwritten Tamil text recognition using the MWDCNN method has achieved an accuracy of 99.3%.