摘要
为提升图像边缘检测效果,更好地捕捉图像中的重要几何特征,根据非下采样轮廓波变换(nonsubsampled contourlet transform,NSCT)系数与图像空域像素位置一一对应的特点,提出一种基于模极大值的NSCT域边缘检测法:首先,在各尺度高频子带中沿着边缘梯度方向提取模值较大的系数;然后,根据累积直方图的值设置双阈值,以抑制图像中较小纹理及噪声产生的系数;再将所检测各方向模极大值系数的位置对应至空域,得到边缘图像。实验结果表明,该方法能检测出丰富的弱小边缘,并能有效抑制噪声的影响。
To improve the image edge detection and better capture the key geometric features in the image,a method of edge detection based on maximum modulus in the nonsubsampled contourlet transform(NSCT)domain is proposed according to the characteristic that NSCT coefficients correspond to the spatial pixel positions in the image.First,coefficients with larger moduli are extracted along the edge gradient direction in high-frequency sub bands at various scales.Next,a dual threshold is set based on the cumulative histogram values to suppress the coefficients generated by smaller textures and noise in the image.Then,the detected maximum modulus coefficients from various directions are mapped back to the spatial domain to obtain the edge image.Experimental results show that this method can detect abundant small edges and can effectively suppress the influence of noise.
作者
石丹
SHI Dan(School of Electronic Information Engineering,Nanjing Vocational Institute of Transportation,Nanjing 211188,China)
出处
《南通职业大学学报》
2025年第2期62-66,共5页
Journal of Nantong Vocational University
基金
2023年度江苏省教育科学“十四五”规划专项课题(C/2023/02/21)
2025—2026年度江苏职业教育研究课题(XHYBLX2025171)
2023年南京交通职业技术学院院级科研项目(JZ2312)
2024年南京交通职业技术学院基本建设项目(JXJG202409)。
关键词
非下采样轮廓波变换
边缘检测
模极大值
nonsubsampled contourlet transform
edge detection
maximum modulus
