Research And Implementation Of Automatic Viscometer Verification Method Based On Image Recognition

Viscosity parameters determination has an important meaning in pharmaceutical,oil,printing,automobile and other industries.Viscometer constant is used to sign viscometers and must be conducted in order to detect whethter viscometer is in line with the standard or not.Capillary viscometer constant can be obtained by a division with sample liquid viscosity value and a in a fixed time that certain amount of the liquid at a constant temperature and completely under the action of gravity through a constant part of a capillary viscometer.Viscometer is needed to be manually checked every once in a while.This kind of method can't overcome the human eye subjective error of reading,also cannot meet the requirements in measurement department of a large number of viscometer verification.Therefor it has the very high innovation and practical significance to find the right image recognition technology with the use of automation equipment to improve the detection accuracy and automation degree while to reduce the error at the same time and to implete the subject line of real-time rapid detection contacting the computer application.This article mainly finished the research and implementation:First,the article used image preprocessing,threshold segmentation and projection technology to segmentate the region of interest.In interest region projection scanning and template matching is used for scale calibration,When detecting camera is stationary the projection scanning method may be used for rapid positioning and when the camera moves the template matching method may be used for accurate positioning,the do the simple stable.Second,the article used a blend of background difference,frame difference and edge detection methods fortarget detection to a nonrigid transparent liquid for its identification and location to improve the detection accuracy and precision.Third,this thesis this thesis designed and implemented the real-time image detection method in order to solve the complex and long running time problem in high precision method that adopted the method of average background modeling and block processing to detect and recognite the target.Chunking can reduce the processing time,was adopted to realize real-time detection.The experimental results show that:scale line detection based on projection scanning and template matching method can effectively identify the calibration,besides target detection method based on edge fusion can pinpoint the liquid level position and has high detection rate,meanwhile block real-time processing detection algorithm also has higher accuracy and faster running speed.