摘要
针对传统检索方式不适用于准规则斑图花型图案检索的问题,本文提出了基于手绘的准规则斑图花型图案检索算法。该算法以基元图案的形式代替原始准规则斑图花型图案,通过阈值分割和Canny算法进行边缘提取,最后基于HOG特征实现相似性度量和检索。实验结果表明,该算法能够较好地检索出与手绘图形状相似的准规则斑图花型图案,检测准确率达83.11%。
Aiming at the problem that the traditional retrieval method is not suitable for the quasi regular pattern retrieval,a pattern retrieval algorithm based on sketch quasi regular pattern is proposed. Firstly,the original quasi regular pattern is replaced by primitive pattern.Then the edge is extracted by threshold segmentation and Canny algorithm. Finally the similarity measurement and retrieval are realized based on hog features. The experimental results show that the algorithm can retrieve the quasi regular pattern which is similar to the sketch pattern,and the detection accuracy is83.11%.
作者
张梦玉
张华熊
ZHANG Mengyu;ZHANG Huaxiong(School of Information Science and Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China)
出处
《智能计算机与应用》
2021年第9期67-70,共4页
Intelligent Computer and Applications
基金
绍兴市"揭榜挂帅"重点攻关项目(2020B41006)
浙江省服装个性化定制协同创新中心项目(浙教高科[2016]63号)。
关键词
准规则斑图
基元分割
花型图案检索
手绘图
quasi regular pattern
primitive segmentation
pattern retrieval
sketch