为了高效地利用数字化书法作品,提出一种基于骨架相似性的书法字检索方法.首先用融合数学形态学的索引表骨架提取方法(MFITS)获取骨架,即先使用索引表对书法字进行初步细化,再用数学形态学的方法细化得到骨架.在提取检索字及候选字骨架后,通过计算和比较检索字与每个候选字骨架的相似性进行书法字检索.与基于轮廓相似度的书法字检索方法相比,文中方法所需检索时间减少了70%,而查全率和查准率基本保持不变,提高了大数据量书法字检索的效率.
To make full use of the digitized calligraphic works, a skeleton similarity-based calligraphic character retrieval approach is proposed. The calligraphic character is firstly skeletonized with morphology fused index table skeletonization (MFITS) method, in which the character is thinned by an index table and then thinned with mathematic morphology method. The similarities between skeletons of the input character and each candidate character are calculated and sorted to retrieve the calligraphic character after they are obtained. Compared with contour similarity-based calligraphic character retrieval method, this method saves 70% of the retrieval time while the ratio between precision and recall are almost the same. It makes the calligraphic character retrieval more efficient and effective.