摘要
针对基于机器视觉的自动拼接碎纸片技术对于不规则碎纸片的拼接存在适应性差、匹配成功率低等问题,提出了一种新的边缘形状描述子方法并与色彩信息相结合的多维度匹配的不规则碎纸片拼接方法:首先,提取碎片边缘的位置信息和颜色信息;然后,基于关键点可有效提高目标匹配效率的特点,将边缘位置信息分段,拟合出边缘曲线,通过边缘曲线得到分段曲线的曲率,获取关键点;最后,在灰度值完成不规则碎纸片拼接的基础上,将关键点的距离信息作为匹配信息,找出相邻碎片信息。拼接完成后通过碎片边缘的彩色信息验证拼接是否正确。该方法有效结合了边缘形状和灰度信息,对不规则碎片的拼接更具适应性;在不同场景下的匹配成功率可达90%以上,并且简化了计算量。
The automatic stitching technology of fragments has an important role in the fields of restoration of cultural relics, restoration of judicial material evidence and medical image processing. Now, the stitching is mainly done manually, and its efficiency is pretty low, so it is particularly important to implement computer automatic stitching technology. Traditional automatic stitching of scrapped paper based on machine vision has poor adaptability and low accuracy for irregular scrapped paper. Aiming to solve these problems, a multi-dimensional stitching method of irregular scrapped paper based on combining edge shape descriptors with color information is proposed. Firstly, the position and color information of the edge of the fragment are extracted.Then, based on the features that key point can effectively improve the efficiency of fragment splicing, edge position information is segmented and the edge curve is fitted, and the key points by the curvature of piecewise curve are obtained.Finally, regard the distance information of key points as matching information to find the adjacent fragments based on using the gray-level to complete the fragments stitching. After the stitching is completed, the color information of the edge of the fragment is used to verify whether the splicing is correct. The innovation of this method lies in the effective combination of edge shape and grayscale information, and it is more suitable for the mosaic of irregular fragments. Compared with the traditional method, this method improves the matching success rate of irregular fragments in different scenarios to more than 90%, simplified calculations and reduce time consumption.
作者
徐睦浩
衣禹桥
李英群
李金屏
XU Muhao;YI Yuqiao;LI Yingqun;LI Jinping(School of Information Science and Engineering,University of Jinan,Jinan 250022,Shandong,China;Shandong Provincial Key Laboratory of Network Based Intelligent Computing(University of Jinan),Jinan 250022,Shandong,China;Shandong College and University Key Laboratory of Information Processing and Cognitive Computing in 13th Five-Year,Jinan 250022,Shandong,China)
出处
《陕西师范大学学报（自然科学版）》
CAS
CSCD
北大核心
2020年第6期90-95,共6页
Journal of Shaanxi Normal University：Natural Science Edition
基金
国家自然科学基金(61701192)
山东省重点研发计划项目(2017CXGC0810)。
关键词
碎纸片拼接
图像处理
形状描述子
曲率
多维度匹配
stitching of scrapped paper
image processing
shape descriptor
curvature
multi-dimensional matching
