Handwritten Tamil Word Pre-Processing and Segmentation Based on NLP Using Deep Learning Techniques

Main Article Content

Dr. R. Kishore Kanna
Iskandar Muda
Dr. S. Ramachandran

Abstract

Tamil is a traditional Indian language spoken mostly among South Indians, SriLankans, as well as Malaysians. This paper proposed the novel techniques based on pre-processing and segmentation of handwritten Tamil words through NLP using threshold value based RGB image conversion to grayscale image. Then to segment this image based on line boundary detection with Alex Net based Convolutional neural network (Alex Net- CNN) in deep learning architecture. Every text is scaled in to needed pixel in the suggested system, that is then exposed to be trained. – i.e., every scaled word contains a set pixel count, which are used to train networks. The findings reveal that proposed method achieved better detection accuracy in written vocabulary knowledge that are equivalent to features extraction techniques. For numerous pictures, a descriptive analysis was performed in terms of effectiveness, accuracy, recollect, and F1 measure.

Article Details

How to Cite
Kanna, D. R. K. ., Muda, I. ., & Ramachandran, D. S. . (2022). Handwritten Tamil Word Pre-Processing and Segmentation Based on NLP Using Deep Learning Techniques. Research Journal of Computer Systems and Engineering, 3(1), 35–42. Retrieved from https://technicaljournals.org/RJCSE/index.php/journal/article/view/39
Section
Articles