摘要
为了解决基于形状匹配的书法字检索计算量大、耗时长、效率低的问题,提出根据样本字特征动态改变剪枝范围的自适应匹配法.离线统计分析数据库中书法字的各特征值分布范围;当用户提交查询样本字后,在线计算查询样本字中各个特征的显著因子,根据不同的显著性因子自适应获取可能相似的候选字集合;利用轮廓形状相似性算法在候选字中进行精确匹配,用匹配值排序检索结果.实验结果表明,与单纯的形状匹配法相比,该方法在提高查全率与查准率的同时,将平均检索时间缩短至5%左右;与层次式匹配法相比,该方法在运行时间上没有明显缩短,平均查全率和查准率提高10%左右.
An adaptive matching algorithm was proposed according to discrimination power of each visual feature in order to overcome the large computing and time-consuming in shape-based calligraphy character matching.Each feature value＇s distribution range was analyzed statistically off-line.When a query was submitted,its features were extracted and the corresponding significance factors were computed based on which self-adaptive algorithm was employed to find a shortened list of possible similar candidates from the database.Then contour shape matching was run on the shortened list to rank and display the similar.The experimental results showed that the adaptive matching approach shortened the retrieval time to 5% of the original shape matching approach.The approach didn＇t significantly speed up the retrieval,but raised the precision ratio about 10% on the condition of the same recall ratio compared with the hierarchical approach.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2016年第4期766-776,共11页
Journal of Zhejiang University：Engineering Science
基金
国家自然科学基金资助项目(61303100)
上海海事大学校基金资助项目(20130467)
关键词
自适应匹配
显著性因子
书法字
adaptive matching
significance factor
Chinese calligraphy
