Research On Several Problems Of Tube-type Bottle Inspection Technology Based On Machine Vision

Machine vision technology, with broad market prospect, is widely used in relevant industries which are difficult to realize by artificial vision work. The purpose and target of this project is to design a scaled, real-time inspection system based on machine version towards some kinds of medical bottles. In this paper, it describes a successful research which realizes automatic inspection towards bottle size and bottle default location and classification according to tube-type bottle features. And it focuses on figuring out several important problems as follow:1. Hardware design. Set down the overall plan of this inspection system and demonstrate its feasibility and practicability, including lighting system and image acquisition system design and selection of hardware devices. At last, introduce experimental situation in the lab.2. Boundary extraction of bottle image obtaining qualified edge image. Describe image processing flow, test various classic image processing algorithms for comparison, and apply improved and innovative algorithms which are more proper for bottles' features such as using wavelet composition and decomposition measure and multi-structure mathematical morphology algorithm as main points in the processing. The comparisons of different measures with this novel algorithm reflect the latter's superiority.3. Relevant inspection algorithms design according to project requirement. Carry out relevant inspections after image processing, concentrating on calculations of matrix pixels and count of geometric parameters of defaults' locations. Introduce the details referred to how to realize these inspected algorithms step to step in compile environment, and estimate probably existent errors in this system adding the improved instructions, and carry out experiments referred to artificial neutral network classification. At last, simply describe programming interface(GUI).