Research On Surface Defects Detection Algorithm Of Precision Injection Molding Products And System Design

As the fast development of manufacturing industry, it has lay claim that products inspection should conform to mass production, high accuracy, quickness and even dynamic testing, which is the traditional human defection cannot meet.Machine vision inspection technology use digital image as the basic information carrier, closely integrated to computer technology, automatic control technology and other subjects, which have the characteristics of non-contact, high precision.Precision injection parts that largely used in electronic products have the high requirement ofappearance quality.The appearance defects of automated vision inspection system are the technical problem that many enterprises need to solve.Due to the complexity of the processing technology, there is high reflective on the surface of precision injection molded products and complex texture characteristic of the optical imaging.At present, there is still no mature and reliable automatic equipment for this special surface defect defection.This paper tooka further study and analysis on the optical characteristics of precision injection parts surface and type of defects, then choose the appropriate light source combination for the system, design a set of machine vision defects automatic defection system that can capture various defects of high reflective surface and detect the defections, then make analysis and improvements ofsome related image processing algorithms that based on a large number of optical imaging experiments.Main work includes:1.Based on the analysis of common way of lighting experiments, design the image acquisition system with a coaxial light source and structured light, which can effective detect the defects of the product. The coaxial light source can detect the dusts, spots, obvious scratches and concave-convex defects, which ensure the test speed of the system; The structured light with adjustable parameters created by software can get precise image of concave-convex defect, which meet the high precision requirement of the system.2.In the aspect of image detection algorithms, the noise that come from the process of image acquisition and image processing can be removed by the methods of filtering and morphology; On the basis of the summarize and analysis of character of image, make improvement ofthreshold segmentationalgorithms by using gradient sharpening filter and introduced the using of sub-pixel accuracy into the improvement of algorithms; On the parts of surface which exist regular texture, take research on the texture feature extraction and defects segmentation; Finally, according to the feature of area, perimeter and the minimum bounding rectangle,study on the classification algorithm.