文章摘要
王瑶,尤丽华,吴静静,宋淑娟.基于改进 SIFT的图像快速自适应匹配算法[J].包装工程,2014,35(11):96-99,104.
WANG Yao,YOU Li-hua,WU Jing-jing,SONG Shu-juan.Fast Adaptive Image Matching Algorithm Based on Improved SIFT[J].Packaging Engineering,2014,35(11):96-99,104.
基于改进 SIFT的图像快速自适应匹配算法
Fast Adaptive Image Matching Algorithm Based on Improved SIFT
投稿时间:2014-03-26  修订日期:2014-06-01
DOI:
中文关键词: 尺度不变特征变换(SIFT)  图像匹配  特征点描述子  图像检测
英文关键词: scale invariant feature transform (SIFT)  image matching  feature point descriptors  image detection
基金项目:国家自然科学基金项目(61305016);江苏省科技支撑项目(BE2011031);江南大学自主科研计划青年基金(JUSRP1059)
作者单位
王瑶 江南大学 无锡 214122 
尤丽华 江南大学 无锡 214122 
吴静静 江南大学 无锡 214122 
宋淑娟 江南大学 无锡 214122 
摘要点击次数:
全文下载次数:
中文摘要:
      目的 分析 SIFT算法特征点描述子生成比较缓慢和匹配过程距离比阈值无法自适应调节的问题, 探索一种改进SIFT的图像快速自适应匹配算法。 方法 通过简化特征点描述子的生成过程, 提高算法效率, 并通过自适应地调节距离比阈值参数, 提高算法的鲁棒性。 结果 与原算法相比, 图像快速自适应匹配算法的匹配点数减少, 匹配准确度高, 计算时间缩短。 结论 该算法在匹配效率和准确度方面具有明显优势, 能够将计算机视觉用于工业无损检测、印品图像检测、印刷网点图像检测和包装图像检测中,具有较好的推广价值。
英文摘要:
      Objective To analyze the problems of very slow generation of feature point descriptors and incapacity of self adaption and adjustment of the distance ration threshold during the matching process in original SIFT algorithm, and to propose a fast adaptive image matching algorithm based on improved SIFT. Methods The efficiency of the algorithm was improved through simplifying the feature point descriptor generation process, and the robustness of the algorithm was improved through adaptive adjustment of the distance ration threshold parameters. Results The test results showed that the algorithm matched more accurately with smaller amount of calculation, though the matched points became fewer. Conclusion With the advantages of efficiency and accuracy, the proposed algorithm was suitable for computer vision in industrial nondestructive detection, printing image detection, printing dot image detection and packaging image detection,with good promotional value.
查看全文   查看/发表评论  下载PDF阅读器
关闭

关于我们 | 联系我们 | 投诉建议 | 隐私保护 | 用户协议

您是第20985285位访问者    渝ICP备15012534号-2

版权所有:《包装工程》编辑部 2014 All Rights Reserved

邮编:400039 电话:023-68795652 Email: designartj@126.com

    

渝公网安备 50010702501716号