Published Paper Details:

Abstract: Segmenting ancient Tamil characters from stone inscriptions presents a significant challenge due to the complex structure of the script and the degraded nature of historical artifacts. Tamil, one of the world's oldest languages, has evolved through various scripts including Sangakala Tamil, Karoshti, Brahmi, Vatteluttu, Grantha, and the modern script, resulting in a rich and diverse character set comprising 247 unique symbols. These characters often contain loops, curves, and holes, making segmentation from stone surfaces particularly difficult. Traditional image processing techniques struggle to address issues such as uneven lighting, surface erosion, and similar foreground-background textures. Accurate character recognition heavily depends on the success of the segmentation phase, which isolates meaningful character regions from noisy backgrounds. This paper presents an in-depth analysis of segmentation techniques applied to Tamil stone inscription images, covering preprocessing methods, segmentation strategies, and recent advancements incorporating deep learning approaches. By emphasizing segmentation as the foundation for reliable character recognition, this study aims to guide future research in developing more robust and accurate Tamil epigraphy analysis systems.