摘要
本文分析了在感知目标与其背景时Otsu判别准则与人类视觉机理间的不一致性,并根据视觉的非线性和适应性原理提出了新的目标图象分割计算模型和算法.实验结果表明,与传统的分割法相比,该方法具有优良的从低反差图象中抽取目标的性能.
It is analysed in this paper that the behavior of the human vision system (HVS) differs from Otsu's discriminant criterion when perceiving an object and its background. 'Based on the visual nonlinearity and adaptability, a new computational model and a new algorithm for object image segmentation are presented. Experiments show that, compared to some traditional algorithms, this method reveals excellent capability of extracting objects from the images with low contrast.
出处
《计算机学报》
EI
CSCD
北大核心
1993年第4期248-256,共9页
Chinese Journal of Computers
基金
国家自然科学基金
关键词
图象分割
视觉非线性
计算机视觉
Image segmentation, visual nonlinearity, machine vision, object recognition.