【摘要】 高分辨率遥感影像具有数据量大、波段少、地物细节纹理信息更加清晰、空间信息更加丰富等特点,因此基于高分辨率遥感影像的道路提取方法研究是当前一个研究热点。但高分辨率遥感影像提供了更丰富的地物目标细节的同时,也使得噪声信息随之增加,如道路上的车辆、道路线、邻近的行树及阴影、建筑物及阴影等,并且在光谱通道上道路与噪声之间存在更强的粘连性。因此,本文通过分析高分辨率遥感影像的道路特征,提出了T型匹配模板道路提取算法,该算法使用角度纹理特征预测道路初始的前进方向,利用T型模板计算道路预测点,采用灰度值最小二乘方法求解最佳道路点,最终提取道路的中心线轨迹。多幅高分辨率遥感影像实验结果分析表明,本文算法可以有效地克服高分辨率遥感影像中道路周边障碍物和遮蔽的噪声影响,完成对道路的有效提取。更多还原

【Abstract】 High-resolution remote sensing images have the characteristics of large amount of data, few bands, more detailed texture information and more abundant spatial information. Therefore, the study of road extraction method from high-resolution remote sensing image is a hotspot. However, high-resolution remote sensing images also increasing the noise information, such as vehicles on the road, road lines, adjacent rows of trees and shadows, buildings and shadows, etc.. There is a stronger adhesion between the road and the noise on the spectral channel. Therefore, this paper analyzes the road feature of high-resolution remote sensing image, and proposes a road extraction algorithm for T-shaped matching template. The algorithm uses the angle texture feature to predict the initial direction of the road, and uses the T-shaped template to calculate the road prediction point. The least square method solves the best road point. Finally, the center line of the road will be extracted. The results show that this algorithm can effectively overcome the influence of obstructions and shielding noise in high-resolution remote sensing images, and complete the effective extraction of the road.更多还原