The observation, decomposition and record of motion are usually accomplished\nthrough artificial means during the process of motion analysis. This method not only\nhas a heavy workload, its efficiency is also very low. To solve this problem, this paper\nproposes a novel method to segment and recognize continuous human motion\nautomatically based on machine vision for mechanical assembly operation. First, the\ncontent-based dynamic key frame extraction technology was utilized to extract key\nframes from video stream, and then automatic segmentation of action was implemented.\nFurther, the SIFT feature points of the region of interest (ROIs) were extracted,\non the basis of which the characteristic vector of the key frame was derived. The\nfeature vector can be used not only to represent the characteristic of motion, but also\nto describe the connection between motion and environment. Finally, the classifier is\nconstructed based on support vector machine (SVM) to classify feature vectors, and\nthe type of therblig is identified according to the classification results. Our approach\nenables robust therblig recognition in challenging situations (such as changing of\nlight intensity, dynamic backgrounds) and allows automatic segmentation of motion\nsequences. Experimental results demonstrate that our approach achieves recognition\nrates of 96.00 % on sample video which captured on the assembly line.
Loading....