Visual inspection of bubble cover packaging machine

Visual inspection of bubble cover packaging machine

There are three ways of drug packaging: bottle, bag and blister. The aluminum-plastic blister packaging of tablets occupies a large share in drug packaging. In the process of tablet production and packaging, it is necessary to go through flow operations such as stirring, drug administration, pressing and sealing, cutting and so on. The traditional manual selection method and weight testing method can not meet the increasingly developing production needs.

Therefore, a set of automatic inspection system for aluminum-plastic blister tablet packaging was developed by using machine vision and automatic control technology. The unqualified products such as vacuoles, imperfections, cracks and foreign bodies were detected in real time and eliminated, which provided favorable technical guarantee for realizing 100% inspection rate (i.e., "zero defect").

Machine vision

Machine vision is a technical discipline that studies the function of human and biological visual system simulated by computer. It uses visual means to obtain the image of the measured object and compare it with the known standard in advance, so as to determine the quality of the measured object. Machine vision technology is widely used in all aspects, almost throughout the entire industrial field, and even can be said to need human vision almost all occasions need machine vision. At present, the international application of visual system is in the ascendant, 1998 market scale is 4.6 billion DOLLARS, and in China, industrial visual system is still in the concept of initiation period, the industry's leading enterprises in solving the problem of production automation, has begun to focus on visual detection automation.

Its research goal is consistent with human vision, that is, let the computer have the ability of sensing, abstraction and judgment on the measured object, so as to achieve the purpose of recognition and understanding. Typical machine vision system should complete the following tasks: image acquisition, preprocessing, feature extraction, recognition and classification, and analysis and description.

Image acquisition is the use of image input equipment (such as camera) to obtain the measured object, and get the corresponding two-dimensional digital image of the visual world.

Image preprocessing is to filter, enhance and correct the two-dimensional digital image to make it consistent with the actual scene as far as possible.

Feature extraction is to analyze and extract the surface color, texture or surface contour of the corresponding object (collectively referred to as feature) on the basis of the pre-processed TWO-DIMENSIONAL image.

Recognition and classification is to analyze and compare the results of feature extraction according to the object "mode" in the pre-existing computer, judge whether the object of interest exists in the image and determine its position. Or distinguish between categories of objects to be measured in identifying images.

Understanding description is to estimate and infer the three-dimensional information of the object from the two-dimensional image, including the three-dimensional space position, the orientation of the surface shape, etc.