摘要
针对SIFT算法对烟包标签图像进行匹配时会有少量错误匹配的问题,提出一种基于改进SIFT算法的烟包标签匹配算法。首先,利用中值滤波方法对烟包标签模板和待匹配的烟包图像进行去噪处理;其次采用SIFT算法提取烟包标签模板和烟包图像的特征点;然后根据余弦相似度进行特征点匹配;最后计算匹配连线的斜率,对斜率结果进行K-means聚类,从而去除错误的匹配结果。实验结果表明,该方法可以解决烟包标签图像匹配结果出现的错误匹配问题。
Aiming at the problem that there will be a small number of mismatches when matching the cigarette packet label image with the SIFT algorithm,a cigarette packet label matching algorithm based on the improved SIFT algorithm is proposed.First,the median filter method is used to pre-denoise the packet label template and the packet image;secondly,the SIFT algorithm is used to extract the feature points of the cigarette packet label template and cigarette image;then match the feature points according to the cosine similarity;finally,calculate the slope of the matching line,and perform K-means clustering on the slope result to remove the wrong matching results.Experimental results show that this method can solve the problem of mismatching in the matching results of cigarette packet label images.
作者
孙显辰
王保云
刘坤香
彭娟
陈美坤
谷雨薇
SUN Xian-chen;WANG Bao-yun;LIU Kun-xiang;PENG Juan;CHEN Mei-kun;GU Yu-wei(School of information,Yunnan Normal University,Kunming 650500,China)
出处
《电脑知识与技术》
2020年第24期187-188,194,共3页
Computer Knowledge and Technology
作者简介
孙显辰(1996-),男,硕士研究生,主要从事泥石流孕灾机制及机器学习研究;通讯作者:王保云(1977-),男,博士,讲师,主要从事机器学习及图像处理研究。
