摘要
针对传统Harris角点检测算法进行改进,提出了一种基于改进Harris的图像拼接算法.首先,通过修改角点响应函数,引进8邻域比较以及圆形非极大值抑制窗口,采用NCC算法对检测出的Harris角点进行粗匹配,采用RANSAC算法剔除误匹配,最后对匹配图像进行图像融合,完成图像拼接.仿真结果表明,该算法提高了图像拼接的准确性、稳定性和鲁棒性,能够达到无缝拼接,具有优良的实用性.
An improved algorithm for image mosaic is proposed based on Harris comer detection. First, the improved algorithm changes the corner response function, calculates the similarity within 8 neighborhoods of the target pixel and makes non-maxima suppression within the circular region. Then, it makes a coarse matching for the detected Harris comer points by using the NCC algorithm. Then, it reduces the the false matched pairs by RANSAC algorithm. At last, the algorithm fuses the spliced images and eliminates the stitching gaps to complete image mosaic. Simulation results show that this improved algorithm increases the accuracy, stability and robustness of image mosaic. In a word, this improved algorithm can reach a target of seamless mosaic and has good practicability.
出处
《南京师范大学学报（工程技术版）》
CAS
2014年第1期70-75,共6页
Journal of Nanjing Normal University(Engineering and Technology Edition)
