快速准确地确定单个样本的所属类别以及总体样本类别数是解决非监督模式识别的前提,然而它们的确定通常是非常困难的。通过研究基于遗传算法的相似性度量最优分类算法以及最优分类数确定算法,提高非监督识别的准确性,并将所研究的算法应用到飞机识别当中。实验结果表明,本算法可以进行最优分类及分类数的确定。 Determining the classification of a single sample quickly and accurately and the overall sample size are the prerequisites for unsupervised pattern recognition, however, their determination is often very difficult. By studying the optimal classification algorithm of similarity measure based on genetic algorithm and the algorithm of determining the optimal classification number, the accuracy of unsupervised identification is improved and the algorithm studied is applied to the aircraft identification. Experimental results show that this algorithm can determine the optimal classification and classification number.