一种改进的鱼眼图像轮廓提取算法

An Improved Approach of Contour Extraction of Fish Eye Image

对鱼眼图像进行轮廓提取是利用鱼眼图像生成球面全景图的前提,鱼眼图像轮廓提取的扫描线逼进算法具有计算量小、速度快的优点,应用较广。针对其存在对噪点抑制能力不足的缺点,提出建立新的扫描线、控制阈值的选取以及对半径进行修正等方法对扫描线逼进算法进行改进,以提高算法对噪点的抑制能力,增强校正畸变能力。实验证明,算法能准确地提取鱼眼图像轮廓,计算量小,实时性好,获得了良好的校正效果。

Contour extraction of fish eye image accurately is the necessary condition of panorama which is generated by fish eye image. The advantages of scan line approach algorithm which is the general method of contour extraction are simplicity in calculations and high efficency. The disadvantage of this method is hard to suppress the noises. In this paper, the algorithm of rebuilding, new scan line, reselecting the threshold and revising the radius are proposed to improve the ability of the scan line approach algorithm in suppressing the noises and calibrating the aberrance. The experiment result shows that the algorithm proposed in this paper can distill the contour of fish eye image more accurately and efficiently.