A binarization algorithm for historical arabic manuscript images using a neutrosophic approach

In this paper, an improved thresholding approach based on neutrosophic sets (NSs) and adaptive thresholding is proposed. This is applied to degraded historical documents imaging and its performance evaluated. The input RGB image is transformed into the NS domain, which is described using three subsets, namely the percentage of truth in a subset, the percentage of indeterminacy in a subset, and the percentage of falsity in a subset. The entropy in NS is employed to evaluate the indeterminacy with a λ-mean operation used to minimize indeterminacy. Finally, the historical document image is binarized using an adaptive thresholding technique. Experimental results demonstrate that the proposed approach is able to select appropriate image thresholds automatically and effectively, while it is shown to be less sensitive to noise and to perform better compared with other binarization algorithms.