A PET Bottle Mouth Detection Method Based on Morphologic

During the production of PET bottles, we must detect the roundness and flaws of the bottle mouth in order to prolong the guarantee period of the drink. After obtaining the images from the product line, we used both common and morphologic method to denoise. Then we extracted the region of interest by enhancing the contrast and binary. We scan from the horizontal and vertical direction to form a grid. And Scan from another two perpendicular axes to form a new grid. We located the centre of the bottle mouth and check whether it is round enough by these multi-direction grids. At last, we used radial scanning to detect the scratches. We calculated the points in each of the sweep intervals and recognized the scratches by choosing a reasonable threshold. Comparing to the core method and Hough Transform method [1], the method to locate the centre of the round is simpler and faster. And we can check the roundness and detect the flaws efficiently. In experiments, we can detect the scratch in size of 0.15mm which can meet the demand of the industrial production.

[1]  Erhard Schubert,et al.  Vision system for empty PET-bottle classification inside the box , 1994, Other Conferences.

[2]  Li Bi Multi-level Fuzzy Enhancement and Edge Extraction of Images , 2000 .

[3]  Wen Tan,et al.  Empty bottle inspector based on machine vision , 2004, Proceedings of 2004 International Conference on Machine Learning and Cybernetics (IEEE Cat. No.04EX826).