论文部分内容阅读
为改善以Hough变换提取遥感影像中弯弧线状地物的准确性,提出了一种新算法M&H(Morphology and Hough),它将Hough变换、数学形态学、非线性拟合等方法串行递进使用,以实现复杂形态线状地物自动提取和直接生成道路矢量图的目标。经适量实例的Matlab仿真测试,证明了M&H的递进噪声去除法优于单次数学形态学降噪法,对含弯弧道路的提取性能也优于标准Hough变换。精度检验结果表明:M&H方法比常规Hough变换提取道路的欠提取率下降了约18.9%。
In order to improve the accuracy of using Hough transform to extract curved linear objects from remote sensing images, a new algorithm, M & H (Morphology and Hough), is proposed. The Hough transform, mathematical morphology, Into the use of complex linear objects in order to achieve the automatic extraction and the direct generation of road vector target. The Matlab simulation test of the proper example proves that the M & H progressive noise removal method is superior to the single mathematical morphological noise reduction method and the extraction performance of the curved road with arc is better than the standard Hough transform. Accuracy test results show that the under-extraction rate of M & H method is about 18.9% lower than that of conventional Hough transform.