Study On Visual Inspection System In The High-speed Automatic Filling Line

The quality safety of food packaging is an important aspect of food safety. Quality inspection of packaging bottles for the foods and drinks before filling is geting more and more attention. With the improvement of the automation degree in filling lines, the current maximum filling speed can reach 20 bottles per second. Manual detection means which is still used in many enterprises has been unable to meet the requirements of high-speed filling lines. The empty bottle inspection system based on machine vision technologies can meet the requirements of high-speed, real-time, and high-accuracy automatic inspection. At present, such equipments used in domestic enterprises are still totally dependent on imports. So it meets the urgent demand of domestic market to research and develop the high speed empty bottle inspection system based on machine vision technologies with independent intellectual property to substitute the expensive import equipment.The inspection system based on machine vision technologies in the automatic filling line is high-speed online inspection equipment which collects machine vision, precise machinery and real-time control. Its main technology includes optics, machine vision, image processing, pattern recognition, information fusion, automatic control and so on.The key technologies researched in this paper about inspection system based on machine vision technologies in the automatic filling line are as following:1. On the principle of the stable performance, simple structure and easy to realize, the paper designs the overall structure of the system and selects the hardware on the basis of analyzing the process of filling.2. This paper studies the key technologies of image acquisition. According to the inspection need and the characteristic of the bottles, the paper chooses hardware equipment which can conform the performance requirement and the reasonable illumination scheme and designs an optical imaging device by multi-layers reflection to get omnidirectional sidewall image of empty bottles. In order to get clear images of high-speed moving bottles, the paper uses the method of LED stroboscopic illumination cooperating with high-speed electronic shutter of CCD.3. The paper presents a high-efficiency automatic tracking algorithm for target detection area in which the pixels will be processed. The other area will not be processed to improve processing speed. The paper presents a method of radial boundary scanning based on judgment under Prior Information to get accurate edge points, and this method can eliminate various interferences. The system uses the method of least-square circle fitting to get the position of image processing area in the process of bottle mouth inspection based on the comparison research on method of locating the center of gravity with method of least-square circle fitting.4. The system uses the method of radius constraint circle fitting to get the position of image processing area in the process of bottle bottom inspection by making the comparison research on method of least-square circle fitting based on chain code constraint with method of least-square circle fitting. The system gets edge points by taking linear scanning in bottle neck and there are two axisymmetric edge points on every line. The paper gets the central axis position of image processing area in the process of bottle sidewall inspection by analyzing special position relation of edge points get from linear scanning.5. The paper selects and extracts reasonable characteristics of connected region according to actual inspection requirements and recognizes the information of bottle images by defect recognition algorithm based on expert decision. The paper investigates the processing method of anti-skid grain on the bottle and presents the recognition algorithm for anti-skid grain on the bottle based on connectivity analysis.