摘要
本文将具有自适应滤波特性的经验模式分解(EMD)方法与Snakes模型相结合,提出了适用于高噪声复杂背景下表面缺陷图像检测的新方法。首先采用改型EMD在整幅图上将缺陷的位置和大致范围从复杂背景下分离出来,然后在缺陷附近用向心搜索法缩小缺陷的检测区域,最后用Snakes模型迭代逼近缺陷边缘。该方法能自动、准确地完成球形金属表面的缺陷检测。实验结果表明,该方法是有效和可靠的。
Combining the empirical mode decomposition (EMD) of self-adaptive filtering with Snakes model,a novel method is proposed to detect surface defect on complicated background image with high noise. Firstly, defect position and its approximate region are separated from complicated background on the whole image using modified EMD. And then, the defect detection region is further reduced using towards-center search method in the defect region. Finally, defect edge is approached in iterative manner using Snakes model. This method was used to carry out the defect image detection on a spherical metal surface automatically and accurately. The experimental results show that the method is much effective and reliable.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2006年第12期1664-1669,共6页
Chinese Journal of Scientific Instrument