基于相位一致性的遥感影像道路特征检测方法

在空间域中进行道路特征检测容易受图像亮度、对比度和噪声的影响,为了克服这一缺点,该文采用在频率域中进行道路特征检测。基于频率域的相位一致性道路特征检测方法具有亮度和对比度不变性,特征在频率域中表现出最大的相位一致性,通过计算影像6个方向的相位一致性值得到相位一致性值之和的灰度图,对灰度图进行二值化,然后使用数学形态学方法进行细化,去除噪声,检测出道路特征。实验表明,该方法对对比度、亮度变化不敏感,能够精确地检测出与人眼视觉相一致的道路特征。

Road Feature Detection Based on Phase Congruency Using Remote Sensing Image

In spatial domain,the results of road feature detection are susceptible to the brightness and contrast of image.To overcome this shortcoming,the features in frequency domain have been detected in this paper.The phase congruency method which is based on frequency domain to detect feature has invariance to brightness and contrast,and features of image are shown to be the largest phase congruency in the frequency domain.The phase congruency value of six directions have been calculated,and then the sum of the six values are obtained,a fixed threshold has been set to get a binary image.The road can be detected by mathematical morphology,and the noise is also removed.Experiment results show that the method is not sensitive to the changes of brightness and contrast,and it can detect road feature accurately,and the features which are detected by phase congruency are consistent with human feature perception.