中文摘要：

在对装配作业人员进行动作分析的过程中,动作的识别和记录一般通过手工操作完成,这种方法不仅工作量大,而且效率低。为解决该问题,提出了一种新的基于机器视觉的装配动作自动分割与识别方法。首先利用基于内容的动态关键帧提取技术提取视频流中的关键帧,实现动作的自动分割;然后提取感兴趣区域的尺度不变局部特征点,据此得出关键帧的特征向量;最后,基于支持向量机构建特征向量分类器对动作进行分类。装配线上样本视频的实验结果表明,所提方法达到了96%的正确识别率。

英文摘要：

The observations, decompositions and records of motions were usually accomplished through artificial means during the processes of motion analyses. This method had a heavy workload, and the efficiency was very low. A novel method was put forward herein to segment and recognize con-tinuous human motions automatically based on machine vision for mechanical assembly operations. First, the content-based dynamic key frame extraction technology was utilized to extract key frames from video stream, and then automatic segmentations of actions were implemented. Further,theSIFT feature points of the region of interested were extracted？ on the basis of which the characteristic vectors of the key frame were derived. Finally, a classifier was constructed based on SVM to classify feature vectors,and the motion types were identified according to the classification results. Experi-mental results demonstrate that the proposed method achieves correct recognition rates of 96 % on sample videos which were captured on the assembly lines.