The Movement Track Detection For Controllable Mechanism Based On Computer Vision

Determining the movement parameters of moving object from an image sequence is an active field of computer vision. Comparing with traditional methods of movement detection, the means based on computer vision can perform lots of tasks beyond traditional ones' scope, because this approach has the merits which can work under non-contact and on-line real-time conditions. This method based on computer vision will find wide application in the future. On the other hand, accurate detecting movement parameters of moving object is necessary for the feedback close loop control of machine based on computer vision.The main task of this paper is to detect the track of controllable linkage mechanism based on computer vision. Firstly, a system platform of computer vision is designed and formed. The operation of image gathering, storage, display and output can be performed on the platform. A newly mended two steps approach is introduced for calibrating the camera system while one order radial distortion exists, which is the most common distortion and has a noticeable effect. The running experimental result shows that the new approach is efficient and accurate. During the course of tracking and gathering, the noises inevitably appear in the image of the object and its background. These noises have disadvantageous effect on speed of recognition and location algorithm, so it's necessary to eliminate these noises. Kinds of image processing algorithms are introduced in this paper. An object detection and sub-pixel location system based on computer vision is described. Fast object detection, accurate segmentation and precise location are realized in the system. Finally, an experiment for the track detection of controllable mechanism based on computer vision is carried out, and a series of tracks has been obtained. The visual detection method for controllable mechanism is realized, and the result is satisfactory.