A New Otsu Thresholding Method on the Circular Histogram
Jiulun Fan, Si Peng · 2022 4th International Conference on Natural Language Processing (ICNLP) · 2022
Using H component of HSI color model to segment the color images is an alternative way. Considering the H component histogram is a typical example of the circular histogram, a new Otsu thresholding version on the circular histogram is proposed based on the theory of circular statistics. Different from the three Otsu thresholding methods on circular histogram defined by Lai and Rosin, the proposed version define the threshold selection criterion by minimizing the angles distance between each point in the class and the mean of the class. The experimental comparison with Lai and Rosin’s three related thresholding methods and the maximum entropy thresholding method on circular histograms shows that the proposed Otsu thresholding version is competitive.