Research Of Hose Surface Flaw Detection System Based On Computer Vision

Computer vision technology is developed by many subjects such as digital image processing and artificial intelligence disciplines. It is used more and more widely in the field of industrial detection. With the rapid development of industry and modernization of production needs, market requires continuously to improve hose quality. Traditional method of detection cannot meet the production requirements, hence the urgent need of a hose manufacturer is the hose surface flaw detection system.Facing the objective demand in the hose market, this paper carries on the thorough research of rubber hose surface flaw image acquisition, the original image processing, feature extraction and defect type discriminant based on computer vision technology. This paper includes the following content:1) The principle of the flaw detection in the hose surface based on computer vision is described. It selects the devices according to the module and sets up detection system to collect the original image.2) Comprehensive utilization of machine vision and image processing technology, automatic defect detection algorithm is proposed. Firstly, It preprocesses the collected image with the improved filtering algorithm so as to eliminate noise and protect the defect edge, then uses edge detection method, morphological dilation processing step to segment the flaw’areas.3) It researches flaw of different forms and defines three main types of flaws. Then it chooses the characteristics of all kinds of flaws pointed and extracts parameters, marks connected area in order to complete defect judgment and classification, which will benefit the identification and statistics of the flaws’ shape and type.The hose surface flaw detection system based on computer vision proposed in this paper has important research value and application prospect. The system can stably do non-contact, automated testing work, and achieve more accurate results, which is better than artificial detection way in terms of both efficiency and accuracy. It is also expected to replace the traditional means of detection and proved to be viable.