The Application Of Machine Vision In Inspection Of Printed Code Defects On Packaging Bags

With the improvement of people's living standard, people pay more and more attention to food health and safety. The quality of food packaging bags plays an important role in the health and safety of food. According to the food packaging industry packaging printing defect detection requirements, the paper designs a packaging printing detection system based on machine vision. This system consists of line light source, image acquisition module, image processing module, the transmission module .It can eliminate the common defects, such as leakage printing, wrong printing, ink contamination and incomplete,etc.The unqualified products will be removed b y using nozzle. In this paper, the key technology involved in character defect detection system is studied deeply.Firstly, this paper introduces the research background and the development of machine vision detection based on character coding defects. Then, according to the characteristics of industrial testing and the actual situation in the industrial field, this paper designs a set of printing character defect detection experiment system. The main contents of this system include three aspects: 1. The design and construction of the system hardware equipment,including machine design and check of the main parts ,such as industrial camera, image acquisition card , light source, external trigger device of camera ,etc. 2.The design and implementation of coding character recognition and defect detection algorithm. 3. The design and implementation of non-conforming reject module. Experiments show that the system can adapt to the actual production needs.This paper studies the character recognition and defect detection from the aspect of image processing algorithms. In this paper, median filtering and threshold segmentation are applied to the image processing.The affine transformation is performed on the pre-processed image. The angle correction of the character region is carried out for the interference of industrial scene instability. In this paper, a normalized correlation coefficient matching method is proposed to realize fast location of character region. In the lattice code character feature extraction stage, the paper repeatedly uses image morphological opening and closing operation to meet a certain distance of grid connected into a whole. Through the analysis of the connected features of printing each character, this paper puts forward a kind of character segmentation method. It realizes the adhesion correct segmentation of the characters. Image recognition module in this system is the core of the recognition method with artificial neural network classifier, and the reliability of the system algorithm is verified by experiment.Finally, this paper develops a set of visualization software for Coding character defect detection using VC ++ development environment. The whole planning of the inspection software adopts the modular idea. The system includes the image acquisition module, the general setting module, the image processing module and the rejecting module , etc.