In order to solve the difficulty of low visibility PM2. 5 hazy weather images edge detection, an edge detection method using nonsubsampled contourlet transform (NSCT) is proposed. Firstly, the PM2. 5 hazy weather image is decomposed by NSCT. Then through calculating the difference of two fine scale coefficients decomposed by the same coarse scale, get the maximum modulus points which can be regard as the edge points. Finally, the integrity binary edge image is obtained by edge fusing on the same scale and between different scales. The results of the experiment show that the proposed method has the advantages of edge integrity, positioning accuracy and fewer false edge points.%针对PM2.5雾霾图像能见度低、边缘检测困难的问题,提出了一种采用非下采样轮廓波变换(NSCT)的边缘检测方法.首先对PM2.5雾霾图像进行NSCT分解.然后寻找由同一粗尺度系数分解而来的两个细尺度相邻子带系数,对其求差值,再通过差值图的模极大点来确定边缘点.最后通过NSCT域尺度内和尺度间的融合得到完整边缘图像.实验表明,对雾霾图像,该边缘检测方法所获取的边缘完整、定位准确并且噪声点少.
展开▼