Key Technology In The Impurities Sorting System Bases On Image Detection

With the improvement of modern life, glass has become the common building supplies and decorations around people. This means large amounts of waste glass will be produced during people’s activities. A domestic enterprise is committing to waste glass recycling. Yet the key problem of the recycling is how to separate glass fragments and impurities.The method which uses the gray-scale detection of machine vision for sorting glass fragments and impurities is simple and convenient. But as a result of that glass fragments refraction of light can format spots in the pictures, thereby it will make the accuracy rate of simple gray detection not high enough and affect the quality of plate glass which is made by the glass fragments.The spots which produced by glass fragments refraction usually occupy50%of total area of the glass fragments, and which produced by impurities can reach90%. For this reason, by using with image enhancement, edge detection and other technical means, extract the edge of the measured object firstly, then by means of compute the image area ratio that spots occupy and compare with the given confidence interval finally, to reach the purpose of distinguishing impurities and glass fragments, whose accuracy rate can reach up to99%.The paper will focus on image enhancement and edge detection technology affect on glass fragments. Image enhancement can improve image blurring which cause by illumination, and edge detection technology can help for impurity determination. These are mainly reflected in the following two aspects of the work:(1) The common algorithm of image enhancement usually adopts more than two templates for image computing. Large amount of calculation and long processing time make it not suitable for the sorting system. Single-Scale Retinex enhancement algorithm is easy to hardware and software implementation. It can achieve dynamic range compression and sharpening the edge details very well and it’s an adaptive enhancement method. Experimental results show that Single-Scale Retinex enhancement algorithm can improve image quality effectively. The effect accords with the requirement of the research.(2) For edge detection, the LoG operator is simple for calculation, but is very sensitive to noise. Therefore it needs to combine other method to improve the detection effect. And mathematical morphology can smooth noise effectively and easy to hardware and software implementation. So the paper propose a method innovate which combined with LoG operator and mathematical morphology. Experimental results show that the combination of both have a very good detection effect.The paper finally introduces the software implementation of the system and gives a comprehensive assessment for the instantaneity and veracity of the impurity sorting system. It also analysis the shortage which exist in the system and gives the direction of improvement,...