基于最大梯度和灰度相关的两步全景图拼接算法-中国科技论文在线.PDF

Vol. 17 No. 12 系 统 仿 真 学 报 Dec. 2005 JOURNAL OF SYSTEM SIMULATION ·2965 · 基于最大梯度和灰度相关的两步全景图拼接算法 薛 峰，张佑生，江巨浪，偶春生 （合肥工业大学计算机与信息学院, 合肥 230009 ） 摘 要：提出一种基于最大梯度和灰度相关的两步全景图拼接算法。该算法综合了基于灰度相关和 特征相关算法的优点，首先通过匹配梯度最大值点位置的梯度值缩小搜索范围，再利用最大灰度相 关匹配两幅待拼接图像。实验表明，算法原理简单、拼接速度快，同时消除了对有重复纹理特征图 像的伪拼接现象，对图像亮度差异具有很强的抗干扰能力，鲁棒性好。 关键词：全景图；拼接；梯度；灰度相关 文章编号：1004-731X (2005) 12-2965-04 中图分类号：TP391.9 文献标识码：A Two-Pass Mosaic Algorithm of Panoramic Image Using Maximum Grads and Intensity Correlation XUE Feng , ZHANG You-sheng, JIANG Ju-lang, OU Chun-sheng (School of Computer Science and Information, Hefei University of Technology, Hefei 230009, China) Abstract: A two-pass mosaic algorithm of panoramic image using maximum grads and intensity correlation was proposed. The algorithm synthesized methods of intensity matching and feature matching image mosaic methods. First, by matching maximum grads of two images, a search range for successive intensity correlation matching process could be remarkably shrunken which made the whole matching process more speedy. Then intensity correlation matching algorithm was used to search the optimal matching position of two images. These two processes ensure the algorithm simple, fast, robust, and false matching can be avoid. Some experiments demonstrate the simplicity, speediness, and efficiency of the algorithm. Key words: panoramic image; mosaic; grads; intensity correlation 引 言 时，在左边图像L 中选取一个匹配模板T，利用模板 T 在右