【摘要】 和大多数侧重于后处理的算法不同,文中在研究了合成孔径雷达(SAR)图像中道路的灰度及几何特征之后,着重在图像预处理中提出了非线性自适应保边缘滤波及局部均方差加权边缘增强算法,在去除乘性斑点噪声的同时较好地保持了道路边缘特征,再从全局的角度考虑提出了一种改进的随机Hough变换(GRHT)算法来确立道路边缘线,继而进一步确定道路中心线。整个算法可以自动进行,具有很高的处理速度及较低的储存需求,实验证明该算法能够准确有效地提取SAR图像中道路目标。更多还原

【Abstract】 This paper proposesd a technique of non-linear adaptive edge keeping filtering and a technique of enhancing the contrast between roads and circumstance using local variance in pretreated SAR image,which improve the efficiency by reducing speckle and the ability to preserve fine features. An improved Randomized Hough transformation is put forward to establish the side line of road, then to extract the road center line. The result of experiment shows that this method can detect roads accurately and efficiently.更多还原