基于骨架特征的生丝抱合开裂检测算法  

Crack detection algorithm of raw silk cohesion based on skeleton characteristics

作　　者：王佳东 孙卫红 邵铁锋 梁曼[1,2] WANG Jiadong;SUN Weihong;SHAO Tiefeng;LIANG Man(College of Mechanical and Electrical Engineering,China Jiliang University,Hangzhou 310018,Zhejiang,China;Cocoon and Silk Quality Inspection Technology Institute,China Jiliang University,Hangzhou 310018,Zhejiang,China)

机构地区：[1]中国计量大学机电工程学院,浙江杭州310018 [2]中国计量大学茧丝绸质量检测技术研究所,浙江杭州310018

出　　处：《上海纺织科技》2025年第2期84-87,116,共5页Shanghai Textile Science & Technology

摘　　要：为使生丝抱合检验结果更加准确,根据实际检验情况定义茧丝脱落和丝条膨胀两种生丝开裂类型,针对茧丝脱落造成的生丝开裂问题,提出一种基于骨架特征的生丝抱合开裂检测算法。首先,对生丝二值图像采用ZS细化算法获取生丝骨架特征;其次,利用删除模板去除位于图像边界位置的非骨架冗余像素点;然后,通过搜索骨架分叉点和打断骨架分支将茧丝分支从生丝主干上分离,并根据骨架连通域的外接矩形长宽比实现对开裂区域的提取。结果表明,算法的平均识别准确率可达94.4%,具有较高的精度。In order to make the inspection of cohesion results more accurate,two types of raw silk cracking are defined based on the actual inspection situation:cocoon shedding and filament expansion.To address the problem of raw silk cracking caused by cocoon shedding,a crack detection algorithm of raw silk cohesion based on skeleton characteristics is proposed.Firstly,the ZS thinning algorithm is used to obtain the silk skeleton features for the binary image.Secondly,the non-skeleton redundant pixels located at the image boundary are removed by using the deletion template.Then,the silk branches are separated from the raw silk trunk by searching the branching points of the skeleton and breaking the branches of the skeleton,and the cracking region is extracted according to the outer rectangular aspect ratio of the connected domain of the skeleton.The experimental results show that the average recognition accuracy of the proposed algorithm can reach 94.4%,showing a high precision.

关 键 词：生丝抱合 图像细化 骨架特征 开裂检测 算法 

分 类 号：TS141.9[轻工技术与工程—纺织材料与纺织品设计]