融合视觉感知和正则割的二维阈值分割算法

TWO DIMENSIONAL THRESHOLD SEGMENTATION BASED ON VISUAL PERCEPTION AND NCUT

阈值法是图像分割的一种重要方法,其关键是如何确定阈值。提出一种融合视觉感知和正则割的二维阈值分割方法,该方法首先利用视觉感知的特性选择候选阈值向量所在的灰度区域,再将正则割作为准则,从候选阈值向量中选出最佳的分割阈值向量。在一系列图像上的实验结果表明,与几种经典的阈值分割方法相比,所提方法的分割效果更好。

Thresholding is an important means of image segmentation, its key is how to determine the threshold value. In this paper, we present a two-dimensional threshold segmentation method which fuses the visual perception and Ncut. The proposed method first utilises the characteristic of visual perception to select the grayscale region where has the candidate threshold vectors, then it uses Ncut as the criterion to determine the optimal segmentation threshold vector from candidate threshold vectors. The experimental results on a series of image show that the proposed method outperforms some classic thresholding methods in segmentation effect.