Development On The Embedded On-line Detection System For Foreign Matter In Ampoule Liquid

Water injection, ampoules as containers, has been widely used in clinical medical, and most was injected into the body through meridians or intramuscular. If water injection contains small particles, patients'health will be threatened directly or indirectly. The current detection results and efficiency of light inspection is not guaranteed because of men's subjective factors and visual factors, therefore, ampoules on-line automatic detection system is very important theoretical and practical significance to improve production efficiency, to guarantee the quality of the product.In recent years, the machine vision technology has been widely used in industrial areas of automatic detection, which provides a new way for liquid agent foreign.Aiming at on-line detection of. amp bottles of liquid agent foreign problems in testing process and combining with the current situation of the development of the on-line detection system, comprehensive use of embedded technology and image processing technique,an amp bottle injection with agent foreign on-line detection system is designed and developed .This paper introduces machine vision technology applied to the development of on-line detection system of survey and image processing, and the related technologies of the system structure, system selection and the sheme of system are discussed. Based on this work,ARM + CPLD as the core of the ampere bottles of liquid agent on-line detection system foreign hardware platforms is designed. This paper mainly expounds the design idea and implementaion methods of the functional modules in the hardware system platform, and then timing simulation results based on QuartusII 9.0 environment is designed and gived. In system implementation, aimed at the characteristics of large amount of data in image acquisition process, an SRAM ping-pong structure cache two 256K is designed to provide .At the same time, the technology of bus segregation and the related control thechnology is increased, which enhance system operation speed and improve data stability.With the characteristics of water injection, processing images by the CDF biorthogonal wavelet transform and inspecting of suspected foreign matter area preliminarily by frame differential method, then the segmentation image was obtained based on fuzzy Tsallis - Havrda - Charvat entropy goal segmentation, finally,ultimate foreign matter segmentation image was obtained by using weighting method combining Preliminary detection with two division results. Matlab simulation results show that the proposed algorithm is feasibe and effective.Finally, the processing speed of embedded system ampoules on-line detection and the reliability of actually eliminate unqualified products were evaluated. Experiments results show that the system is stable and requirements that the on-line detection processing needed is satisfied.