CAJ | 학술논문

针对矩形文档碎片的拼接提出了一种聚类的碎片边缘灰度匹配算法，这种算法主要涉及碎片的预处理、边缘灰度的提取、碎片的聚类以及匹配，其中碎片聚类和匹配是碎片拼接中最关键的组成部分。碎片聚类可以使计算机搜索待匹配碎片的数量减少，提高匹配的正确率。碎片的匹配算法选取了碎片的左、右（或上、下）两边界的部分边缘灰度值，大大压缩了信息量，它不仅可以提高计算机处理速度，而且匹配精度高、效果好。实验证明，此算法适用于矩形碎片的拼接，具有高效、快速、稳定的特点。
침대구형문당쇄편적병접제출료일충취류적쇄편변연회도필배산법，저충산법주요섭급쇄편적예처리、변연회도적제취、쇄편적취류이급필배，기중쇄편취류화필배시쇄편병접중최관건적조성부분。쇄편취류가이사계산궤수색대필배쇄편적수량감소，제고필배적정학솔。쇄편적필배산법선취료쇄편적좌、우（혹상、하）량변계적부분변연회도치，대대압축료신식량，타불부가이제고계산궤처리속도，이차필배정도고、효과호。실험증명，차산법괄용우구형쇄편적병접，구유고효、쾌속、은정적특점。
Proposed an algorithm of boundary-grayscale matching based on cluster for rectangle paper fragments.It is mainly related to the pretreatment of debris,the edge extraction of gray-value,fragments of clustering and matching.Clustering and matching is the most key part of the fragments stitching.By clustering,the number of the matched pieces can be reduced to improve the matching accuracy.It is advantageous to the computer sear-ch.Moreover,the matching algorithm selects parts of the left or right(or upper and lower)boundary-grayscale, which greatly compress the amount of information.It can not only improve the computer processing speed but also have high matching precision and good effect.Experiments showed that this algorithm is applicable to match the rectangle pieces.Furthermore,it has the characteristics of efficient,rapid and stable.