论文部分内容阅读
本文研究基于SOM(Self-Organizing Feature Map)神经网络学习模型的高分辨率遥感影像道路网自动提取算法。首先利用数学形态学提取遥感图像道路的初始道路区域信息,自动对原始图像进行分区并确定神经元初始权值,用SOM网络学习模型对神经元进行训练学习,经迭代获取道路网中心点位置,最后运用“中心点四邻域跟踪判别法”跟踪连接形成道路中心线。实验表明,该方法在高分辨率遥感影像道路网的提取上有较好的效果,特别在主干道路网的提取上效果更佳,对噪声干扰具有良好的鲁棒性。
This paper studies the automatic extraction algorithm of road network of high resolution remote sensing image based on SOM (Self-Organizing Feature Map) neural network learning model. First of all, mathematical morphology was used to extract the initial road area information of remote sensing images road, automatically partition the original image and determine the initial weight of neurons, SOM network learning model for neurons training and learning, iteratively obtain the location of the center of the road network, Finally, using the “four-point center-point tracking method” tracking connection to form the road center line. Experiments show that this method has good effect on the extraction of road network of high resolution remote sensing images, especially on the extraction of the main road network, and has good robustness against noise interference.