Multilingual Off-Line Handwriting Recognition in Real-World Images Using Adaptive Neuro Fuzzy Inference System (ANFIS)
Handwriting has been used as one form of communication and information which is recorded in day-to-day human’s life. Among this multiple handwritten recognition available in machine, it is significant field where recognition mechanism has been applied. During this study, English and Tamil languages are provided as inputs which are handwriting languages. Additionally it also made study on how this content could be transfer in to binary data. This review aims to focus on concepts which are behind the algorithms of recognition in off-line strategy. Off-line text could be available in images that are scanned. Since handwritten recognition consists of various stages, segmentation is considered as an important process. The existing system would affect the script rate during recognition by directly doing separation on words, characters or lines. Thus it arises the issues if samples in dataset have same content then finding out similar styles for various users and leads to difficult process. In order to solve this issues clustering process are made over the words and characters with similarity style then recognition has been performed on those clustered output. Denoising of image involves many steps namely noise elimination, binarization, size elimination and thresholding. The segmentation of word is carrying out via bat mechanism. At last, the recognition on words has been made using which stands for Adaptive Neuro Fuzzy Inference System (ANFIS). The results are measured using the metrics like precision, recall, f-measure, accuracy and classification time.
Author's Name: M. Sivasankari, Dr.P. Velmani and Dr.P. Arokia Jansi Rani