摘要
全景图拼接是基于图像绘制的关键技术之一.本文提出了一种基于熵的全景图拼接算法,它利用现有的基于小波的角点提取算法获得兴趣点,同时保留其所在的边缘信息,并对图像局部熵作改进,以此来获得候选匹配点,然后去除伪匹配进而获得最终精确匹配.文中给出了实验结果和算法的鲁棒性分析.
Panoramic image mosaicing is the key technology of IBR. A panoramic image mosaicing method based on entropy is proposed in this paper, it use an existing wavelet-based corner detection algorithm to obtain interesting points, at the same time, it keep edge information around the points. Then an improvement for image local entropy is made using a weighted function, so as to get pseudo matches. In the end, an optimization method is used to discard mismatches, and accurate matches are obtained. The result of the experiment and the robustness of algorithm are also discussed in this paper.
出处
《小型微型计算机系统》
CSCD
北大核心
2007年第1期148-150,共3页
Journal of Chinese Computer Systems
基金
辽宁省教育厅高等学校科学研究项目资助
沈阳市应用开发研究计划项目资助.
关键词
全景图
鲁棒拼接
熵
panoramic image
robust matching
entropy
