针对最大似然法在拒绝类选取上的复杂性，本文给出了高维情况下提取拒绝类的新思路。其作法是直接从高维空间的类条件概率密度函数出发，对各个类别的条件概率密度函数值进行门限界定，从而得到门限判别。在最大似然法的程序实现中，采用了最小距离法提供先验概率。实验结果表明，此改进是有效的，并具有一定的推广价值。 In view of the complexity of maximum likelihood method in rejecting class selection, this paper gives a new idea of ​​extracting reject class in high dimension. The approach is based on the conditional conditional probability density function of high-dimensional space, the threshold of the conditional probability density function of each category is defined, and the threshold is obtained. In the procedure of maximum likelihood method, the minimum distance method is used to provide the prior probability. Experimental results show that this improvement is effective and has some promotional value.