Research And Implementation Of Roping Packaging Breakage Detection Based On Industrial Machine Vision

Against the backdrop of social advance, rising consumption level and fierce market competition, products’ package and quality become the dominant factor in seizing market share. More and more manufacturers recognize how packages would contribute to enterprise efficiency. However, how to conduct package damage test with high accuracy and efficiency remains unresolved. To address the high cost of manual test and false or error detections resulted from fatigue, we turn to computers and machine vision for a way out. Currently many industries are beginning to use a system based on machine vision inspection system to solve the damage test problem. The system has high efficiency, high accuracy and can be in working condition for long time. It can be able to adapt to the characteristics of complex industrial environment, such that the damage test method based on machine vision is increasing popular. However, because the object we reseach in this paper has some unique features of its own, It increases the difficulty of system algorithm. Currently, the market does not have a sophisticated automated equipment can be detected and removed. Currently, the market does not have a sophisticated automated equipment to do package test of rope group.The main work in this paper is to study the image processing part of roping packaging breakage detection system. This paper first introduces the research status and development trend of the machine vision technology and image processing theory. Then we introduce the image process about the roping packaging damaged test from several typical aspects of machine vision systems, including image preprocessing, image segmentation, feature extraction and image classification and recognition. Image pre-processing is the basis of roping packaging breakage detection system. This paper introduce a method make a true color bitmap to convert grayscale bitmap, Image denoising methods and Image enhancement processing method. In view of our object, By analyzing the characteristics of several different algorithms, We select the appropriate image processing algorithms to provide a reliable guarantee for image segmentation, feature extraction, and classification.Image segmentation and extraction is the basis of image analysis. In this paper, several methods were used for image segmentation,we have used the threshold segmentation method、the adaptive threshold segmentation method and region growing segmentation to split image. Compareing and analyze the performance of each algorithm. In the method of classification,we select the BP neural network classifier to classify roping packaging damaged. The non-linear classifier combines the advantages of multi-threshold segmentation method to make up for the impact of environmental change on the algorithm,promoting the robustness of the algorithm. The result shows that the BP neural network classifier can completed the roping packaging damaged test and relatively good.