低能见度下的地平线检测算法研究

Method of skyline detection for poor visibility conditions

为了在一定程度上增强飞行员的环境意识,地平线检测在合成视觉系统中起着非常重要的作用。针对红外图像对比度低、噪声大的特征,提出了一种低能见度下的地平线检测算法。该算法对图像进行灰度变换以增强对比度,根据人类视觉系统识别过程中的注意和捆绑两个阶段以及边缘检测的相关理论,定义了每个像素的能量,利用动态规划法找到候选的地平线分界点,最后经过Hough变换得到真正的地平线。实验结果表明了该算法的有效性和优越性,这在航空领域具有非常重要的意义。

In order to enhance the environmental awareness of pilots to some extent, skyline detection plays a very important role in Synthetic Vision System. Based on the characters of infrared image, which has bad contrast and heavy noises, an algorithm of skyline detection for poor visibility conditions is proposed. The intensity values of the image is adjusted at first, and then the energy of each pixel is defined according to the two phases of human visual system, attention and binding, and the related theory of edge detecting is defined. The alternative skyline pixels with dynamic programming algorithm are gotten. At last, Hough transform is dealt with, and the real skyline is obtained. Comparative study reveals its superiority and robustness, which is very beneficial in the aviation areas.