CAJ | 학술논문

笔迹鉴别是一种有效的身份鉴别方法。作者针对维吾尔文文字书写特点提出了一种文本无关的笔迹鉴别方法，该方法采用递推方式，首先，以边缘图像为基础，利用8邻域像素点，计算边缘像素点四维方上的方向比重；然后，获取微切分窗口内所有的边缘像素点的方向比重；最后，求得窗口中书写笔迹的方向比重，并以此作为特征结构进行笔迹鉴别。实验不但改进了笔迹图像预处理，而且根据特征结构的特点，在加权卡方距离度量度量方法上进行笔迹密度再加权计算。结果表明，该方法提高了鉴别效果，有很强的实用性。
필적감별시일충유효적신빈감별방법。작자침대유오이문문자서사특점제출료일충문본무관적필적감별방법，해방법채용체추방식，수선，이변연도상위기출，이용8린역상소점，계산변연상소점사유방상적방향비중；연후，획취미절분창구내소유적변연상소점적방향비중；최후，구득창구중서사필적적방향비중，병이차작위특정결구진행필적감별。실험불단개진료필적도상예처리，이차근거특정결구적특점，재가권잡방거리도량도량방법상진행필적밀도재가권계산。결과표명，해방법제고료감별효과，유흔강적실용성。
Handwriting identification is an effective method of identification.According to the features of Uygur letters, this paper proposed a text-independent handwriting identification technique. Based on edge image, the author used a recursive system. Firstly, we counted the edge pixel’s 8-neighborhood pixels to get its proportion of four di-mensional directions. Secondly, did this process on all the edge pixels of local windows.At last, obtained the hand-writing’s proportion of directions which we made as feature structures.The method made improvement in image’s pre-preoccessing. According to the the characteristics of feature structures,the author also used the density of stroke to weight weighted Chi-square distance formula again to measure the similarity between handwritings. With a strong practicability, the method achieve a better identification result.