Abstract: In this paper，the hand completeness checking algorithm used in driver physical examination system is studied.Firstly，self-adaptive thresholding algorithm is used after that a RGB image is converted to the YCbCr color space.Then the hand area image is obtained by using the method of gray scale projection from the preprocessed image.Finally，extracted moment feature of the the hand image is used as the training sample data of SVM，and then the classifier is utilized in hand completeness recognition.The hand completeness checking algorithm has high precision.Testing results of the algorithm shows that the method is effecitve and it has been successfully used in the driver physical examination system.

Key words: hand completeness checking, YCbCr color space, Hu momoment, Support Vector Machine（SVM）

摘要： 为实现对手完整性的自动检测，研究了用于机动车驾驶员体检系统中的手完整性检测算法。首先把RGB图像转换到YCbCr色彩空间后对Cb通道图像进行自适应阈值分割；然后对图像采用投影法获取准确的手部图像；利用提取的手部图像Hu矩特征值作为样本数据训练SVM模型，最后利用训练好的分类器对手进行完整性识别。手完整性检测算法可以达到理想的识别精度。算法的测试结果表明对手完整性检测算法是有效的，已被成功应用于机动车驾驶员体检系统。

关键词: 手完整性检测, YCbCr色彩空间, Hu矩, 支持向量机