摘要
提出一种基于高分辨率遥感图像的分水岭分割方法。首先,利用Canny算法进行边缘检测,得到边缘梯度图。其次,利用数学形态学对梯度图像进行后处理。然后,根据自动标记分水岭变换方法实现高分辨率遥感图像分割。最后,利用基于信息熵评价方法,对遥感图像的分割结果进行非监督评价。结果表明:所提方法可以有效地抑制遥感图像的过分割现象,并取得较好的分割效果。
This paper proposes an algorithm of watershed image segmentation on high spatial resolution remote sensing image.First,phase congruency model is used to obtain the edge strength.Second, mathematical morphology is used for post image processing.Third,watershed transform is used to complete the high spatial resolution remote sensing image segmentation.Finally,information entropy is adopted to evaluate the image segmentation results.The evaluation result shows that our proposed method can be used to reduce the over segmentation,and obtain the better segmentation result.
出处
《黑龙江工程学院学报》
CAS
2015年第3期16-18,共3页
Journal of Heilongjiang Institute of Technology
