摘要
常规边缘检测方法对边缘检测不够细腻且受噪声影响,通过引入保边去噪技术并修正广义希尔伯特变换(GHT)边缘检测方法,建立了新的边缘检测方法。该方法首先对资料进行保边去噪预处理,再优选时窗进行频率域加窗傅里叶变换,利用频率域加窗傅里叶变换虚部绝对值沿频率求和的1/n次幂进行边缘检测。模型和实际资料的计算结果表明,该边缘检测方法不仅有更高的边缘检测精度,而且具有抑制噪声能力。该方法可以有效地预测构造裂缝、溶蚀孔洞等缝洞发育带的分布特征。
Conventional edge detection methods have low resolution and serious noise influence.Therefore we propose a new edge detection approach by introducing 2-D edge-preserved smoothing and generalized Hilbert transform(GHT)edge prediction.First edge-preserved smoothing processing is carried out.And then optimized windowed Fourier transform is conducted.Finally imaginary part absolute values of the transform are used to detect edges along 1/N power of the frequency sum of the windowed Fourier transform.This approach is applied to effectively predict the distribution of structural fracture and Karst pores and caves.Model and real data test results show that the proposed edge detection approach has not only strong noisesuppressed capacity but also high precision.
出处
《石油地球物理勘探》
EI
CSCD
北大核心
2015年第3期490-494,5,共5页
Oil Geophysical Prospecting
基金
国家自然科学基金(41104073
41364004
41304097)
江西省自然科学基金(2010GQS0002)
"863"课题(2012AA09A404)资助
关键词
广义希尔伯特变换
边缘检测
保边去噪
generalized Hilbert transform,edge detection,edge-preserved smoothing
