高分辨率遥感影像中飞机识别研究

在目标识别领域,飞机是一种常被研究的对象,对飞机进行精确识别并定位,无论在军事打击和商业调查中都有重要意义。高分辨率图像中丰富的空间、光谱和上下文信息在带来更多有用细节信息的同时,也给飞机识别带来了更加复杂的背景干扰。主要研究应用面向对象的理念来实现对高分辨率遥感图像中飞机的识别。首先是图像预处理,即获得图像中的对象;然后是面向对象的图像识别,即提取图像对象的特征;并依据特征完成识别任务。利用的基本特征主要有图像对象的面积、长宽比、分散度、凸包缺陷。实验证明这种识别方法,对于不同朝向和大小的飞机都能取得较高的检出率。

The Research on Plane Target Recognition in the High Resolution Remote Sensing Image

The plane as a target is always used to study the field of target recognition. Accurately identification and location of the plane has important significance in both the military and commercial investigation. The rich space, spectrum and context information in high resolution image provides more useful information, but it brings about more complex background interference. In this paper, we applied object-oriented method to identify the plane in the high resolution remote sensing images. First step, image preprocessing, that is obtaining the objects in the images; Second step, the object-oriented image recognition, that is getting the characteristic of image object and accomplishing recognition. The characteristic of image include area, length-width ratio, dispersity, ridge defect. The detection rate of the recognition algorithm is high in the experiment, although the planes are different in size and direction.