基于AMSS方程的道路影像去噪方法

Road-Image Denoising Based on AMSS Equation

对道路影像进行去噪处理是道路提取(识别)的重要前提步骤,处理过程中不仅对滤波器的去噪性能有较高的要求,同时对其保持边界的性能要求更高。AMSS(Affine Morphological Scale Space)方程是唯一同时满足对比不变性和仿射不变性的方程。文中从AMSS方程的理论基础着手,从数学角度深入分析了用该方程进行图像处理时能够使图像在去除噪声和保留边缘信息之间达到平衡的内在机理。然后,通过对用于道路提取的遥感影像特征的分析,研究了AMSS方程在遥感影像道路处理中的应用。理论分析与试验表明,文中提出的方法具有良好的效果。

Removing noise from road-image is often the first step in road extraction. The technique to be used should not only reduce the noise, but do so without blurring the edges. AMSS equation is the unique one satisfying all classical invariance properties, as well as projection invariance. Theory foundation of AMSS equation is given at the beginning. In the following, the virtue of its good balance between noise removing and edge preserving are analyzed. The equation is used in road-image processing in the end. The theory and experimental results indicate that this method is effective.