Vision-based Intelligent Inspection Robot Technology For Health Wine

China has a large alcoholic drinks market and their production increased quicklyyear by year, however the automation level of the equipment, especially theautomatic detection, in the alcoholic drinks production line needs improving. Inrecent years, with the increase of consumption and perfection of laws and regulations,the product quality is more and more important in the alcoholic drinks industry. Thenaked-eye detection, which is not efficient and accurate enough, is being challengedby the advanced alcoholic drinks production equipments from foreign countries.Therefore, it is significant to research on alcoholic detection equipment withproprietary intellectual property right in China.The project is designed to develop an intelligent robot to detect the products inthe large-scale and super-speed health liquor automatic production line. Based onmulti-sensor information fusion detection and recognition, the intelligent robot candetect the products automatically and separate the unqualified products from theproduction line. Meanwhile, the intelligent robot can be applied in other fields, suchas drinks and medicine industries, and insure the safety of products in such fields.Health liquor intelligent visual detection robot involves technologies in manydisciplines, such as machine vision, machine design and transmission, electricalcontrol, sensor detection and information integration, optical imaging, digital imageprocessing, communication and so on. This paper, firstly, analyzes the overall designproposal and the technologies that the intelligent robot involves. It secondly presentsthe proved algorithm aimed at detection of visible foreign matter in the health liquorand the bottle detection by researching on the mechanical structure, electrical controlsystem, optical imaging, and visual information acquisition. Finally, this paperpresents a prototype of the health liquor intelligent visual detection robot and appliesthe technology into the production of some medicine, such as transfusion and oralliquid. All in all, the innovative achievements presented in this paper are as follows.1. This paper elaborates the present situation of alcoholic drinks detection,points out the significance of developing an intelligent visual detection robot,introduces machine vision technology, analyzes the reasons why the visible foreignmatter in the health liquor and the bottle defect occur, and focuses on the presentsituation of liquor in China and foreign countries. 2. This paper proves the feasibility of the health liquor intelligent visualdetection robot system, presents the overall technical proposal, compares three kindsof mechanical structure of health liquor intelligent visual detection robot: linearintermittence, taking pictures in a fixed position and rotated tracking, researchesdistributed electrical control structure based on IPC and PLC, analyzes three stagesphotographed by tracking: acceleration, revolve at the same angular speed, rapidreturn, designs the optical imaging that involves detection posts of black foreignmatter, capacity, white foreign matter and bottle cap, presents the robot’s detectionprocedures, compares many kinds of unqualified products sorting methods andanalyzes robot’s software framework.3. This paper analyzes kinematically the visible foreign matter in the healthliquor, compares visible foreign matter, air bubbles, scratches on the bottle walls, putforwards an improved weighted filter algorithm on the basis of analysis of basicfiltering algorithm, promotes an improved adaptive-filtering algorithm based on theextremum and mean value, which can not only filter efficiently the noise, but alsoremain the moving targets, by researching on the weighted average filteringalgorithm based on mid-value, proposes empirical rule positioning method to positionimages based on edge statistics, which achieves good positioning effect.4. This paper researches on method of segmentation and identification of visibleforeign matter in the sequential pictures of health liquor, analyzes method ofextraction of moving target with background subtraction, researches on thresholdsegmentation based on Fisher evaluation function, analyzes moving target trackingmethod based on improved Mean Shift and Kalman filtering. This paper also presentsthe living examples to analyze and test. Based on the shape of the moving target,such as length, width, rectangular degree and degree of circularity, as well asestimation of movement, this paper presents support vectors, researches on movingtarget identification on the basis of support vectors, conducts image calibration. Thenanalyzing from the pressure pixel, this paper presents the measurement of visibleforeign matter and the judgment whether the foreign matter size is within the allowedlimits. Finally, this paper presents the feasibility of visible foreign mattersegmentation and identification algorithm through experiments.5. This paper researches on liquor bottle detection algorithm. Aimed at thedefect that the bottle cape inclines at one side, this paper proposes positioning in theedge special point based on center of gravity and multiple scanning. Directing atdefect that the bottle cap may emboss, this paper promotes a positioning method based on multiple searching. It also researches on detection method of bottle capdefect pictures quickly matching based on layering once the positioning point iscertain. Aimed at the advantages and disadvantages of Sobel operator edge detection,this paper presents an improved Sobel operator to acquire the liquid capacity line andrealize the capacity detection on the basis of liquid surface height judged byexperience.6. This paper presents the prototype of health liquor intelligent visual detectionrobot and promotes its application. It shows the picture of the robot and set-updiagram of the key components, analyzes the software constitution from four aspectsthat is from human-computer interface, testing interface, control software to databasemodule. It finally analyzes Knapp-Kushner detection method of health liquorintelligent visual detection robot, gives test data and applies the technology used tomake health liquor intelligent visual detection robot into transfusion visual detectionrobot, ampoule visual detection robot and oral liquid visual detection robot, whichwere experimented on the spot.We have solved many a problems in theory and in application during the processof researching on health liquor intelligent visual detection robot, from project designto detection experiment, from image pre-processing and positioning algorithm tovisible foreign matter detection and identification and research on bottle detectionmethod, from machine design to electrical control and transmission, from prototyperesearch to its application. All of these provide valuable experience for theequipment’s following upgrade and development; provide the relative equipmentresearch and development with good reference and guarantee the high quality andsafety of the alcoholic drinks and medicines in the automatic production lines.