Research On The Algorithms Of The Machine-Vision-Based On-line Measurement And System Implementation

With the development of auto detective technology, the machine vision, as a newly-developed subject, has been greatly developed because of its efficiency, direct, high accuracy, real-time, simple operation, etc. Especially in the area of on-line quality detective, the machine vision technology is used wider and wider. In this dissertation, an application of this tech in the field of real-time sign distinction on the laser film machine is depicted.This document will introduce an entire machine vision system which is applied in practice of real-time quality-detection from hardware and algorithm aspects. The system will include: the choice of lamp-house, the design of the clamp, the installation of trigger system, the debugging of image grabbing setting, the real information based image processing, boundary detection, object segmentation, calculate of invariant moments, the design of human-computer interface, and so on.Respecting hardware, this dissertation will start from the practice and introduce the reasonable image grabbing equipment. The equipment satisfy the desire of the integrated laser film machine is designed, and solved troubles of illumination, shield of noise and getting rid of vibration efficiently, etc.From software, mathematical model of noise blur and their solution are contrived. The Laplacian is optimized on the aspects of boundary detection and goals recognize, meanwhile some other algorithm are advanced, all these work are for the sake of increasing the overall efficiency.Last but not the least, based on optimized algorithm and practice, the system provided a friendly human-computer interface which provide a series of optional parameter to make the system steady enough to adapt varies situation.This machine vision system is running on the laser film machine in the Baoding SRT factory and Shanghai Zijiang Metallization Packaging Material CO., Ltd with a favorable effect.