基于多标记像素匹配的高分辨率遥感图像道路提取

针对高分辨率遥感图像道路提取方法存在提取不完整和误提取问题，提出一种基于多标记像素匹配的高分辨率遥感图像道路提取方法. 将待提取图像由RGB颜色空间转换到 Lab颜色空间，选取与照明强度弱相关的色相特征作为初始匹配项. 以矩形框的方式标记不同类型的道路，利用t 检验法剔除其匹配项的异常值，从而确定阈值来匹配道路像素点，利用局部纹理算子对匹配结果进行筛选. 利用道路区域的形态特征对匹配结果进行优化处理. 为验证所提方法的可行性和优越性，对不同传感器获取的高分辨遥感图像进行测试，与现有道路提取方法进行对比. 定性和定量精度评价结果表明，所提方法对不同类型道路的提取具有较高的精度.

High-resolution Remote Sensing Image Road Extraction Based on Multi-mark Pixel Matching

Aiming at the problem of incomplete and mis- extraction of road extraction methods for highresolution remote sensing image, a high-resolution remote sensing image road extraction method based on multimark pixel matching is proposed. The image to be extracted is converted from RGB color space to Lab color space, and the hue feature that is weakly related to the illumination intensity is selected as the initial matching term. Different type of roads is marked by drawing rectangular boxes, and the outliers of the matching items are eliminated by t-test, so as to determine the threshold to match the road pixels, and the matching results are filtered by local texture operator. The matching results are optimized by using the morphological features of the road region. In order to verify the feasibility and superiority of the proposed method, the high-resolution remote sensing images obtained by different sensors were tested, and compared with the existing road extraction methods. The qualitative and quantitative accuracy evaluation results show that the proposed method has higher accuracy for different types of road extraction.