AI Based Tamil Palm Leaf Character Recognition

An extensive study aimed at analyzing characters from old Tamil manuscripts written on palm leaves corresponds to Tamil Palm Leaf Character Recognition. These culturally and historically valuable manuscripts require extensive processing in order to effectively extract characters because they are frequently affected by chronological and degradation. The process begins with the capturing of a clear photograph of the manuscript, which is then preprocessed to remove imperfections like stains and smudges. Before preprocessing, the digitalized manuscripts are cropped vertically into three parts to improve the preprocessing quality. To remove color differences and compress the data, the image is converted to grayscale. subsequently noise is removed using specialized filtering algorithms. The image is refined through further morphological processes, which fill in small gaps and blurs character borders. Binarization makes it easier to distinguish the characters from the surroundings. Line segmentation is applied to the previously divided image to separate out individual lines of text for improved character recognition. Finally, the preprocessed scripts are trained by using CNN model. In the proposed model the Tamil characters are effectively recognized with an accuracy of 91%, and resulting in the digital preservation and comprehension of these precious palm leaf written works.