CAJ | 학술논문

首先介绍了一种基于指纹方向图分区和遗传算法的连续分类方案。根据各种典型指纹类别的拓扑特性而引进了一组动态分区模板和代价函数用于引导方向图分区，并将寻找最佳匹配划分的过程转化成一个N维向量空间的函数优化问题，然后给出了该问题基于遗传算法的求解过程。最后，讨论了遗传算法的优越性。
수선개소료일충기우지문방향도분구화유전산법적련속분류방안。근거각충전형지문유별적탁복특성이인진료일조동태분구모판화대개함수용우인도방향도분구，병장심조최가필배화분적과정전화성일개N유향량공간적함수우화문제，연후급출료해문제기우유전산법적구해과정。최후，토론료유전산법적우월성。
In this work, we introduce a fingerprint continuousclassification schema based on drectional image partitioning and GA. A set of dynamic partition masks, according to the fingerprint topology, and a cost estimating function are designed to guide the partitioning procedure. Finding best fitted mask application is converted to an functional optimizing problem, and we give out a GA solution to the problem. At last, we discuss the advantage of GA comparing with the exhaustive search method.