This paper presented an improved image segmentation algorithm based on 2D Otsu, in which two-dimensional histogram was mainly build by gray and neighborhood gray gradient of two tuples, and then calculated the biggest Otsu value of the object area and the background area in image, at last, got the threshold of the image. The experimental results indicate that we can get more smooth area of the image and more accurate shape of the image’s edge, compared with Otsu method and the traditional 2D Otsu method.
Xue-guang,CHEN Shu-hong, WANG
"An improved image segmentation algorithm based on two-dimensional Otsu method,"
Information Sciences Letters: Vol. 1
, Article 2.
Available at: https://digitalcommons.aaru.edu.jo/isl/vol1/iss3/2