Computer vision has being applied widely at national economy, science research and national defense. The object, which just like industrial automation, product line, vision guided navigation and virtual reality etc, is measured no touch is the maximal merit. Computer vision observes a range of the human eye below detection, such as infrared, microwave, ultrasonic etc, computer vision just supersede labors test method step by step, which's strongpoint is high efficiency, fast rapidity and no vision tiredness. But computer vision technology still stays to the moment of consummating and developing.The research aimed at the design of a tube-type medicinal Terrarium image inspecting system based on machine vision. Overall system relates to image processing, DSP technology, it can inspect the defective tube-type medicinal Terrarium with other auxiliary equipment. Thesis mainly aims at research on pivotal technique contacting compactly with tube-type medicinal Terrarium inspection:Firstly, the structure of the inspecting system was introduced. Aiming at the characteristic of tube-type medicinal Terrarium inspection, the Thesis discusses the project as a whole, the selection of acquisition equipment and designing of illumination apparatus. Initial system design objectives have been achieved.Secondly, research on the detection algorithm of the medicinal Terrarium. So after introducing domestic and overseas research actuality about detection and designing programs for some representative arithmetic, comparing large of number of the result, this thesis brings a Hybrid Filtering Method to detect Terrarium defects.Thirdly, the dissertation introduced the real-time image processing system. Processing algorithms were realized on SEED-VPM642, and the high efficiency and feasibility of TMS320DM642 processor has been proved. Using DSP was an advanced method of improving efficiency of the inspection system. |