Noise Removal on Batak Toba Handwritten Script using Artificial Neural Network

This paper proposes the use of Artificial Neural Network (ANN) to remove background noise of Batak Toba handwritten script. Several artificial backgrounds are fused with the original script to reconstruct visual perception of manuscript text background. Experiments have been conducted on offline handwritten script cleaning process and show superior impact compared to threshold method and Gaussian filter. This is a seminal work on ancient Batak Toba manuscript recognition system as the primary research theme.