Machine Vision-based Information Collection System For Collector Coin Grading Boxes

At present,the packaging boxes of collector coins circulating in the market are printed with corresponding grading information.With the development of the collector coin industry,the types of collector coins are gradually increasing,and the traditional manual image acquisition and information extraction methods are difficult to realize batch operation and ensure the accuracy of character information recognition.To address the above problems,with the goal of automating collector coin information collection,a collector coin information collection robot control system was designed,combining machine vision,digital image processing and motion control technologies.In order to realize the fully automatic collection function,a set of collection process flow of coin packaging box was designed,including absorption,box delivery,positioning,turnover,image acquisition and return,it can solve the problem of low efficiency and low quality of manual collection.The communication protocol of CAN bus was established to realize the efficient data transmission between upper and lower computers and ensure the stable operation of the whole control system.Based on the above process flow,a coin collection information acquisition robot based on machine vision was designed,including XY axis moving device,Z axis vertical moving suction device,moving turnover device,positioning moving device and image acquisition device,the structural design of each device was reasonable and compact.At the same time,the selection of stepping motor,motion controller,industrial camera,optical lens and light source required by the robot was completed,and an efficient and stable acquisition platform was built.According to the characteristics of the image of the coin box,the character information recognition algorithm of the coin box was designed: firstly,through a series of preprocessing methods,including grayscale,image enhancement,position correction,binarization and filter noise reduction,the character information area was extracted,and the factors irrelevant to the recognition target in the image were removed,so as to obtain the grayscale image containing high-quality and clear character information;then,the improved projection character segmentation method was used to solve the problems of character adhesion and Chinese character segmentation in the image.Firstly,the string was segmented by line through the horizontal projection method,and then the character width threshold was introduced to segment each line of string into a single character,which effectively solved the problems of missing segmentation of adhesive characters and over segmentation of Chinese characters by the traditional projection segmentation method;finally,the segmented characters were trained and recognized by MLP classifier.The application software of the collection coin information collection robot control system was developed,including motion control module,image capturing module,image processing module and motor test module.Finally,the whole machine test experiments were carried out.The experimental results show that: the collection coin information collection robot control system ran smoothly,image acquisition was clear.The character information recognition rate was up to 99.5%.The collector coin grading box information can be gathered automatically and efficiently through the robot in this paper.