基于机器视觉的钞纸缺陷检测研究-计算机应用专业论文.docxVIP

中国科学院硕士学位论文：基于机器视觉的钞纸缺陷检锢l研究Abstract 中国科学院硕士学位论文：基于机器视觉的钞纸缺陷检锢l研究 Abstract The machine vision technology，which can extraordinarily reduce the inspection cost,effectively ensure the quail锣ofproducts，strongly raise the speed and efficiency of production，has become one hot topic in the industrial online inspection in recent years and is widely used in the field of industrial inspection．While showing a great many merits both in research and application field，the technology also meets some application problems and difficulties when applied to high-speed online inspection task because of the restrictions of large data amount and complicated operation of vision information． In the thesis，觚automatic inspection system，together with its design process and realization scheme iS proposed．The system is based 0B machine vision and is used to detect defects ofbanknote papers on the high·speed product line． Firstly，the design of the automatic inspection system，which includes the functions and architecture design，is thoroughly studied and described according to the requirement analysis ofthe system． And then the design and implement of relative image processing algorithms are discussed in detail，、Ⅳith the requirements of high-speed online inspection strictly and detailedly considered．The algorithms involve ima夸e collimation，grayscale correction， template constitution and defects grading．For instance，a fast grayscale correction method using the optical performance ofthe sheet is proposed in the thesis；In order to overcome the low precision of image collimation，the traditional single-template detection technology is improved and two templates are built to detect low grayscale defects and hi曲grayscale ones separately．A multiresolution defect analysis algorithm based on pyramid decomposition is also designed for the inspection system SO as to achieve fast classification of banknote papers．In the simulated experiments， 0111scheme achieved a good result of being able to detect and grade a