摘要
针对图像中经常存在的斑状特征,提出一种位置与尺寸自动检测方法。首先基于斑状特征的梯度分布构造能够用于该特征检测的极值能量函数;然后从理论上分析所构造极值能量函数的极值特性,并基于模拟图像进行极值特性直观分析;最后给出基于极值能量函数的图像斑状特征位置与尺寸检测实现算法。实验结果表明,本文方法不仅能够有效准确地检测出图像斑状特征的位置与尺寸,而且对图像噪声、模糊、视角变化具有较强的稳定性与鲁棒性。
Focusing on blob features that usually appear in images, we developed a method for detecting their positions and size automatically. The main work includes: 1 ) An extreme energy function for detecting blob features, which is constructed based on the gradient distribution of blob features; 2) A theoretical analysis made on the property of extreme energy function, and then an intuitive analysis of its extreme property is made based on simulated images; 3 ) The implementation of the algorithm for detecting positions and sizes of blob features is proposed ; 4) Experiment results showing that the method proposed in this paper can effectively and exactly detects positions and sizes of Nob features in images, and the method performs stable and robust under noise, image blur, and viewpoint changes.
出处
《中国图象图形学报》
CSCD
北大核心
2012年第5期656-664,共9页
Journal of Image and Graphics
基金
国家自然科学基金项目(61005033
61075033)
模式识别国家重点实验室开放基金项目(20090018)
关键词
斑状特征
位置与尺寸检测
极值能量函数
3维极值能量空间
blob feature
position and size detection
extreme energy function
three-dimensional extreme energy space