论文部分内容阅读
高分辨率卫星遥感技术具有在更小的空间尺度上探测地表目标的能力,利用其影像数据进行车辆检测已成为新的研究热点。在概述遥感影像车辆检测研究现状的基础上,对车辆目标影像特征及车辆检测过程进行了探讨;将车辆检测方法分为利用光谱/几何结构特征的基本检测方法和综合运用多种特征的智能化检测方法,并详细叙述了多种车辆检测方法的原理与适用性以及车辆提取中的关键技术。通过分析发现:结合多特征的机器学习和面向对象的车辆检测方法更适合较复杂环境下的车辆检测。
High-resolution satellite remote sensing technology has the ability to detect surface targets in a smaller spatial scale, and the use of its image data for vehicle detection has become a new research hotspot. Based on the overview of vehicle detection research status of remote sensing images, vehicle target image features and vehicle detection process are discussed. The vehicle detection methods are divided into the basic detection methods using the features of the spectrum / geometric structure and the intelligent application of various features Detection methods, and described in detail the principle and applicability of a variety of vehicle detection methods and vehicle extraction of key technologies. Through the analysis, it is found that the combination of multi-feature machine learning and object-oriented vehicle detection method is more suitable for vehicle detection in more complex environment.