CAJ | 학술논문

针对碎纸机三种碎纸模式进行拼接复原，提出了一种基于动态聚类的文档碎纸片自动拼接算法，定义了匹配度矩阵计算两块碎片最合理的拼接方式，设计了一种基于碎纸片特征向量的动态聚类行聚类算法进行行初步聚类，根据文字特征线及计算出的行距对初步聚类进行了调整修正，确定最终的行分类及行间顺序，根据提出的动态四邻近匹配算法，匹配出复原结果。实验表明，该方法实现简单，成功率高，能快速得到碎纸片的三种碎纸模式的拼接复原结果。
침대쇄지궤삼충쇄지모식진행병접복원，제출료일충기우동태취류적문당쇄지편자동병접산법，정의료필배도구진계산량괴쇄편최합리적병접방식，설계료일충기우쇄지편특정향량적동태취류행취류산법진행행초보취류，근거문자특정선급계산출적행거대초보취류진행료조정수정，학정최종적행분류급행간순서，근거제출적동태사린근필배산법，필배출복원결과。실험표명，해방법실현간단，성공솔고，능쾌속득도쇄지편적삼충쇄지모식적병접복원결과。
This paper proposes an automatic documents fragment re-assembly algorithm based on dynamic clustering. It defines the matching matrix and calculates two fragments most reasonable splicing. It designs a dynamic line clustering algorithm based on eigenvector of torn pieces for a preliminary clustering. According to the text characteristic line and spacing, it adjusts the results of preliminary clustering, which can determine the final line classification and order. It matches the recovery results according to dynamic four adjacent matching algorithm. Experiment results show that this method is simple, and success rate is high, and can get stitching recovery results quickly.