中文摘要：

为了解决传统可拓模式分类器只适合用于模式类别较少和一维数据等缺点，针对可拓学中可拓距和关联函数的特点，提出了采用可拓距和关联函数新的可拓模式分类器，包括平均关联函数法、K最大关联函数法和最大关联函数法。详细介绍三种可拓模式分类器的设计步骤；并针对平均样本法、平均距离法、K近邻法、最近邻法和三种可拓模式分类器，采用二维平面数据、多维向量数据和手指静脉图像特征矩阵数据进行实验对比分析。实验结果表明：新的基于可拓距与关联函数的可拓模式分类器算法不仅可以解决模式类别较多的多维数据问题，分类效果也可以达到经典的K近邻分类器或最近邻分类器的分类水平。

英文摘要：

ABSTRACT： In order to overcome some defects that the traditional extension pattern classifier is only suitable for lit- tle small pattern classification and one - dimensional data, according to the characteristics of the extension distance and the correlation function of extension, three new algorithms of extension pattern classifier were proposed. The al- gorithms are the average correlation function, K - maximum correlation function and maximum correlation function based on the extension distance and the extension correlation function. The steps of the extension pattern classifiers were introduced in detail. For these classifiers of the average sample, the average distance, the k - nearest neighbor, the nearest neighbor and the extension pattern classifiers, the two -dimensional data, the multi -dimensional vector data and the matrix data of finger vein image feature were used to compare and analyze the classification effects of the extension pattern classifiers. Experimental results show that the new extension pattern classifiers can deal with multi- ple patterns and multi - dimensional data, and its classification result can reach the level of k - nearest neighbor clas- sifier or nearest neighbor classifier.