Research Of Mold Detection System Based On Machine Vision

In most of the injection mold industry, usually using artificial to monitor the injectionmold. When the abnormal situation occurs, locking the injection mold machine to preventthe press mold and the generation of defective products, to protect the mold. In injectionindustry field, the quality of detection needs to be improved because of the great laborintensity and the risk. Now the industry want to save labor, developing a mold detectionsystems to achieve injection mold automation detection for all of the problems.In this paper, designing injection mold detection based on machine vision technology. Itsets up a detection system, the system intake site mold images by a CCD camera and sendthe images to the PC to dispose, then sending control signals based on the result of thedetection, so completing the work of the detection. The main tasks: the harsh environmentof dust、smoke、lighting uneven will led a variety of industrial immediately noise, as well asthe mechanical jitter caused image motion blur degradation phenomenon. Selecting imagepre-processing functions like Roberts edge detection definition evaluation operator in orderto estimate the direction and scale of a fuzzy image motion. Wiener filter restore thedegraded image, median filter remove image noise. The pre-image preprocessing finished.To compare the advantages and disadvantages of detection method based on gray templatematching and geometric contours information after the image analysis. Then after the imageanalysis, using geometric contours to locate and match based on the match result todetermine whether the product is qualified by IO output, combined the alarm system toachieve injection mold detection automatically.The injection mold detection system’s development, using image processing softwareHexsight to realize image acquisition, analysis, processing and storage. Using VC6.0todesign interface based on the Hexsight, complete detection system’s design.