Abstract

The complete extraction of roads from remote sensing images (RSIs) is an emergent area of research. It is an interesting topic as it involves diverse procedures for detecting roads. The detection of roads using high-resolution-satellite-images (HRSi) is challenging because of the occurrence of several types of noise such as bridges, vehicles, and crossing lines, etc. The extraction of the correct road network is crucial due to its broad range of applications such as transportation, map updating, navigation, and generating maps. Therefore our paper concentrates on understanding the cognitive processes, reasoning, and knowledge used by the analyst through visual cognition while performing the task of road detection from HRSi. The novel process is performed emulating human cognition within cognitive task analysis which is carried out in five different stages. The suggested cognitive procedure for road extraction is validated with the fifteen HRSi of four different land cover patterns specifically developed-sub-urban (DSUr), developed-urban (DUr), emerging-sub-urban (ESUr), and emerging-urban (EUr). The experimental results and the comparative assessment prove the impact of the presented cognitive method.

摘要

从遥感图像（RSI）中完全提取道路是一个新兴的研究领域。这是一个有趣的话题，因为它涉及检测道路的各种程序。使用高分辨率卫星图像 (HRSi) 检测道路具有挑战性，因为会出现多种类型的噪声，例如桥梁、车辆和交叉线等。正确的道路网络的提取至关重要，因为它具有广泛的应用，例如交通、地图更新、导航和生成地图。因此，我们的论文专注于理解分析师在执行 HRSi 道路检测任务时通过视觉认知使用的认知过程、推理和知识。新过程是在认知任务分析中模拟人类认知进行的，认知任务分析分五个不同阶段进行。建议的道路提取认知程序通过四种不同土地覆盖模式的 15 个 HRSi 进行验证，特别是发达城市郊区 (DSUr)、发达城市 (DUr)、新兴郊区 (ESUR) 和新兴城市 (欧元）。实验结果和比较评估证明了所提出的认知方法的影响。