Research On Moving Object Recognition And Matching Of The Crane Hoisting System

In the industry, bridge crane bears a large number of lifting jobs, which plays an important role inthe industrial development. Due to worse light, a lot of dust and complicated ground condition in itswork environment, some safety accidents often happen. In order to effectively reduce the occurrenceof safety accidents, this paper puts forward a method “Obstacle avoidance and warning system basedon the binocular stereo vision of the crane hosting”, which assists bridge crane driver and improvesthe safety of industrial production.For corporate workshop crane lifting a safe, efficient and stable requirements, design based onbinocular stereo vision technology crane lifting visual warning system. Crane lifting cargo andobstacles for accurate positioning, auxiliary crane drivers safe drive. Meanwhile, the design of thecrane visual simulation platform, crane and identification simulate actual working process of thevisual system. In the image recognition detecting and processing, through the HSV color spacesegmentation and canny edge detection of the edge of the hook and the obstacle markers to determinethe centroid, Calibrate the camera to capture images and region-based stereo matching. Because basedon disparity or gray-based motion detection algorithms are not well identified accurately target.Improved proposes a combination of motion parallax and grayscale target detection, so that the visualcrane lifting system can be external factors such as light or shadows adverse environmental impacts, isstill able to accurately identify the moving objects, and can well target contour extracted. Liftingthrough the MFC software to determine a safe distance between objects and obstacles, and make theappropriate warnings, to achieve the vision system crane lifting Obstacle warning alarm.In the experimental design and research process, a detailed analysis of the visual system warningcrane lifting each sensor installation and selection, Proposed lifting crane visual warning systemarchitecture design methods and sensor structure parameter calculation step, and designedexperimental simulation platform. Through the detection of moving object recognition algorithm isproposed combining parallax and grayscale target detection algorithm to improve the detectionaccuracy of the identification system. Design methods and algorithms are verified by experiment thefeasibility. The result for the practical application has some guiding significance.