Restoration of Taikhar Rock Cultural Heritage Text

Abstract

Rock art is among the most valuable cultural relics, yet it remains vulnerable to destruction by human actions. Preserving cultural heritage inscriptions in their original form is crucial for safeguarding historical monuments. Image processing techniques have proven effective for restoring cultural heritage images. In this research, the R-CNN deep learning method was employed to detect inscriptions on the rocks of Taikhar, located in Arkhangai Province, Mongolia. Additionally, efforts were made to classify authentic cultural heritage inscriptions and spurious texts. Given the rarity of preserved original cultural heritage texts, restoring known inscriptions is a priority. As the natural rock of Taikhar is reddish-brown, color restoration and image enhancement techniques were applied to achieve meaningful results. After restoring the colors of the cultural inscriptions, further restoration methods can be implemented. To this end, an experiment was conducted using the DStretch plugin to enhance the images, yielding improved clarity and accuracy in restoration.