A Study Of Plug Wire Robot Visual Positioning Algorithm

Manual detection method is always uesd to detect telecommunications networks intelecommunications rooms. It determines the locations and reasons of faults by analizingthe returned signals gained by pluging in test terminals to the corresponding jacks ofnetwork patch panel module. It is not only a heavy workload, but has a low efficiency.Often faults can not be fixed for the first time, as a result a robot that can detect networkfaults is required. The robot, however, can not accurately move to the corresponding jacksdirectly, because of the irregularly arranged network patch panels. Thus precise targeting ofthe robot is required.A method based on image processing technology for precise targeting is proposed inthis paper. First, convert the obtained color image into gray scale one, and a new gray scaletransformation formula which well improves the contrast of the image is proposed,according to the differences of three color components RGB between the target jacks andneighborhood interferences. Second, enhance the image by the combination of histogramand gray stretching method. Next segment the enhanced image parallelly using the iterativethreshold method and Otsu threshold method.Third, a template matching algorithm for theshape characteristics of the target is designed. Later, two positioning results are obtained byprecisely locating the two segmented images. And if the difference of the two results isgreater than a threshold, which is determined according to the accuracy of actual projects,we believe that the positioning results are wrong. In order to avoid damaging plug wireterminals, we prompt operators to manually locate faults. If the difference is less than thethreshold, we can increase the positioning accuracy by averaging gained results. And a largenumber of experimental results show that the method can not only protect plug wireterminals, which meets the needs of the actual projects.