Abstract: This paper introduces a new method of gait representation and an improved Bayesian classification algorithm based on it at the same time.A triangle is applied to simulate the legs of key frame which has the maximal or minimal step，and extracts its parameters as the gait feature.KNN and a new N-best algorithm are used for test-data search the optimal numerical value from training data before Bayesian classification.The experimental results on the NLRP database show that the new method is not only simple and fast but also effective.

摘要： 提出了一种基于步态序列中腿部三角特征的步态表示方法，在这种特征上用改进的朴素贝叶斯分类方法进行步态识别。选取步幅最大、最小两种情况下的姿态作为关键帧，用三角型模拟其腿部特征，提取三角型模型参数作为步态特征，识别时先分别用KNN和一种改进的N-best取得属性值在训练数据中的对应数值，然后用贝叶斯分类方法识别。在NLPR数据库上使用留一校验方法进行算法验证，实验证明该方法简单快速，而且取得了比较理想的识别效果。