Development Of Visual Inspection System For The Inner Parts Of Milk Formula Cans And Research On Its Key Technologies

As a special part of the food packaging containers, milk formula cans’quality requirements become increasingly high with the general concern of theparents. However, foreign materials (such as paint, particles, dirt, etc.) will inevitablydrop onto the inner wall or bottom of the can during the production process. Oncethese foreign materials are not discovered, and mixed with the milk formula in thefilling process，a long time latter, chemical change will happen and endanger people’shealth.For the automatic inspection of the defects in the inner wall or bottom of milkformula cans, a visual inspection system for the defects in the inner parts of milkformula cans was developed according to the actual production environment in oneenterprise, and some key technologies of visual inspection system were researched inthe paper. The main contents include:(1) According to the technical specifications and requirements of the enterpriseon the inspection system, overall program and workflow of the visual inspectionsystem of milk formula cans were designed.(2) The hardware structure of visual inspection system for inner parts of milkformula cans was completed: according to the inspection environment andrequirements, the most appropriate device including light source, camera, lens andimage acquisition card was selected to ensure a clear image acquisition; thecorresponding positioning and control systems were also designed in order to get thereal-time image of the milk formula cans and remove the fail one accurately.(3) Algorithm research and system software design: Image filtering and imageenhancement were took to improve image quality, and highlighting the image features.According to the software design, the system software was written in LabVIEW technology, and the main modules and functions of the software system wasdescribed in three parts: parameter settings, inspection and monitoring, datamanagement.(4) Ambient lighting in the inspection process is one of the key technologies ofthe visual detection system. By testing and comparing, a bowl-shaped LED lightsource was chosen in the way of scattered lighting to solve the problem of strongreflecting of the detected material.(5) The algorithm research of accurate positioning and defects detection for themilk formula cans is another one of the key technologies. The method of least squaresfitting was used to achieve the circle positioning of the can. On this basis, thetraditional Hough transform was improved to achieve the weld location. So theaccurate positioning of the can was completed while a basis for the determination ofthe detection area was established. Then according to the defects’ features andcharacteristics in the inner wall of the milk formula cans, the defects types can bedivided into point defects, line defects and surface defects. After using theclassification method in conjunction with the Blob algorithm, the wall defects can beaccurately identified and marked.(6) A static experiments was be carried out in the laboratory for the inspectionsystem, and the result showed that the algorithm for milk formula cans bodydeformation detection is good, while still have some false detection. After beingdebugged, the equipment is now installed to the milk cans production line, and thelive dynamic test shows that the stability and accuracy of the visual inspection systemfor the inner parts of milk formula cans is fine, and the identification accuracy rate is99.70%. So the system we developed can meet the requirements of milk formulacans’ online quality inspection.