【摘要】 提出一种对高分辨率的灰度低空航拍图像的房屋主轮廓识别系统,应用图像分割与房屋矢量化相结合的算法,首先把图像分割成一系列只包含1幢或几幢房屋的子图像,再分别对子图像进行边缘检测、直线检测和房屋矢量化,以缩小算法的复杂度,并实现自适应地调整矢量化参数.该系统适用于房屋密度大,结构复杂的城市地区,在图像平面与地面有一定的倾斜角度时,仍能很好地识别出房屋的主轮廓特征,算法对阴影也有很好的抗干扰性.更多还原

【Abstract】 A recognition system for building’s principal contour from graylevel aerial image of high resolution was proposed. With the combination of image segmentation and building vectorization, the system segments the image into a series of subimages containing one or several buildings, and then does edge detection, line detection and building vectorization respectively. Thus, it not only decreases the algorithm’s complexity, but also achieves an adaptive method to adjust parameters in vectorization automatically. The system can be used in those images taken from the urban areas with buildings of high density and complex structures, and can work fairly well in oblique views and with shadow interference.更多还原