科研之友
微信
新浪微博
Facebook
分享链接
摘要
When using 2-D thresholding to segment images, the used threshold would partition the 2-D histogram into four quadrants, two of which are corresponding to the object and background, while the other two are corresponding to edges and noise. However, unsuccessful segmentation will often happen because many existing 2-D thresholding methods ignore edges and noise quadrants in calculation. To solve this problem, in this paper a novel 2-D threshold line segmentation strategy is proposed, in which the second threshold point is determined adaptively by considering the information of incorrectly classified pixels. The experiments on typical images demonstrated that the proposed method achieves very competitive segmentation results in comparison with the existing representative methods.
