Image Enhancement and Thresholding for Ancient Inscriptions in Trowulan Museum's Collection Mojokerto, Indonesia

Inscriptions are an important source of information that supports archaeological and historical research. However, several significant challenges are due to surface damage, fading, and environmental wear, which often obscure the legibility of inscriptions and interfere with the identification process. This research aims to improve the readability of ancient inscriptions in the Trowulan Museum collection, in Mojokerto, Indonesia, by applying several image processing techniques. Grayscale and Contrast Limited Adaptive Histogram Equalization (CLAHE) methods are applied to improve image quality. Next, four types of thresholding methods including Niblack, Otsu, Sauvola, and Phansalkar were compared to produce the clearest image. Based on the results of calculating the mean, standard deviation, SNR, and PSNR, this research shows that the sauvola method produces brighter and clearer images with the least noise, but has the potential to lose writing identified as noise. On the other hand, the Phansalkar method, with a mean of 200.536, a standard deviation of 104.509, an SNR value of 1.585 dB, and a PSNR value of 6.188 dB, emerges as a viable alternative for preserving fine details in text that are often identified as noise by other methods.