论文部分内容阅读
针对非合作区域遥感图像中复杂背景下车辆目标检测困难的问题,提出一种改进形态学重建的车辆目标检测方法。首先,利用遥感图像的近红外和红色波段的信息,获取场景的归一化植被指数(NDVI),去除植被背景的干扰;其次,设计方向模板,对全色图像进行滤波,使人造物背景得以完整保留,并生成标记图像,利用标记图像重建得到人造物背景;最后,利用原图和重建背景之间的差异,进行车辆检测,有效消除了复杂背景的干扰。将本文提出的方法应用于实际遥感图像的车辆目标检测,结果表明,检测效果好,鲁棒性强,无需先验信息,可用于大幅面遥感图像的车辆目标自动提取。
Aiming at the difficulty of vehicle target detection in complex background in non-cooperative remote sensing images, a modified method of vehicle target detection based on morphological reconstruction is proposed. First of all, the normalized vegetation index (NDVI) of the scene is obtained by using the information of the near-infrared and red bands of the remote sensing image to remove the interference of the vegetation background. Secondly, the direction template is designed to filter the full-color image to make the artificial background to be Complete preservation and generation of mark images, and use of mark images to reconstruct the man-made backgrounds; and finally, using the differences between the original image and the reconstructed background to detect vehicles, effectively eliminating the interference of complex backgrounds. The method proposed in this paper is applied to the detection of vehicle targets in real remote sensing images. The results show that the proposed method has good detection effect and robustness without prior information and can be used to automatically extract vehicle targets in large format remote sensing images.