Research On Technology Of Visual Detection For Visual Foreign Matter In Irregular-shaped Bottled Solution

China is a big country for production and consumption of alcoholic drink and otherbeverage. From2009to2011, continuous statistics on www.askci.com show: The productionof soft drinks and zymolytic alcohol was in a rapid growing stage, the growth rate is up to20%in2011. And the most important thing is it will keep growing in following days. Visualinspection is an important branch of automatical science, which combines optical, mechanical,electronics, and computer hardware and software technology. It is related to computer imageprocessing, pattern recognition, artificial intelligence, signal processing, opticalelectromechanical integration and other areas. The article takes the irregular-shaped bottled125ml Jing Brand health wine as the object of research. This article detailed the keytechnology of the visual inspection of visible foreign matter in irregular-shaped bottledsolution from the inspection robot’s mechanical, electrical, imaging system structure, digitalimage processing, on-line detection and recognition and software structure aspects.Firstly, we have taken a market research on beverage consumption and quality testingstatus quo in China. And on this basis, a brief introduction of technology status quo onmachine vision system and background of visible foreign matter detection. Its contentincludes the technology of light sensation, illumination in machine vision system and themethod of image processing for foreign matter detection.Secondly, a brief introduction on the robot’s sub system and a geometrical opticsanalysis for the photographic system are showed in the second chapter. The result of analysisshowed that there is no imaging blind area and the robot’s imaging system is reasonablypracticable.The algorithms of image processing and moving target detection are the most criticaltechnology in the entire machine vision system. In this paper, we proposed a complete set ofvisual inspection algorithm while considering the mechanical vibration and the dynamic andstatic interference in images. First, identify the bottles in sequential images so that we canfind the system error in lighting great change and generate prompt message and system errorlog. Second, an accurate targeting is used to locate the target bottle and generate the bottlecoordinate system in images. It provides the basic information in the following small movingtarget extraction. Third, on the basis of a large number of literature reading and experimentalstudy, this paper proposed a new method of visual detection based on region-growing andblobs-deleting creatively. The experimental results proved its efficiency, anti-jammingcapability. Finally, a highly efficient software system based on the detection method described inthis article was designed and realized. The software system architecture is divided into fourparts: main detecting program package, camera configuration program package, robot controlprogram package, inspection records management procedures. This software system has threecharacteristic:1. it can deploy cameras easily;2. multi-threading technology makes softwaresystem can run in a highly parallel state, so that the system can process fast, at the same time,each camera corresponding disjoint;3. records management program to ensure the traceabilityof the system equipment test results.