基于预检测SUSAN算法的图像拼接  

Image mosaic based on SUSAN method of pre-detection

作　　者：曲天伟[1] 陈晓丹[2] 曹雪栋[1] 陈桂兰[2] 杨泽雪[1] 

机构地区：[1]黑龙江工程学院计算机科学与技术系,哈尔滨150050 [2]黑龙江科技大学计算机信息与工程学院,哈尔滨150027

出　　处：《信息技术》2014年第2期148-150,155,共4页Information Technology

基　　金：黑龙江省教育厅科学技术研究项目(12511447)

摘　　要：为了快速精确提取图像角点,提出了一种基于预检测的SUSAN角点提取方法。文中首先采用预检测的方法排除部分伪角点,对预备角点采用SUSAN算法检测出图像的角点。然后采用图像分块技术和近邻角点排除方法来保证角点分布均匀,没有成簇现象,保证后续图像拼接的精度。实验结果表明,此方法不但能提高角点的检测速度,而且能合理地提取图像中的角点,提高角点的检测精度,进而提高图像拼接的速度和精度。In order to quickly and accurately extract image comers, A SUSAN comer detection method with pre-detection is proposed in this paper. At first, a corner pre-detection is used to extract candidate comer and reject false comers. Then, image block and neighboring point eliminating method is proposed to make the corners be well-proportioned distributing and avoid too many comers stay together effectively. The experiment shows that the method not only improve the speed of comer detection, but also improve the detection accuracy of the corners with reasonable extraction to the corners. It improve the image mosaic speed and accuracy.

关 键 词：角点提取 预检测角点 图像分块 图像拼接 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]