Defect detection algorithm for high speed inspection in machine vision

Machine vision (MV) is a technique used to provide image based automatic inspection and analysis for applications such as inspection and process control. This paper aims to implement a real time defect detection algorithm for high speed inspection using machine vision for a Mild Steel (MS) welding chip. The performance of the existing defect detection algorithms available for MS welding chips fails due to various noises like thermal noise and digitizing noise. The proposed algorithm aims to reduce the noise efficiently by pre-filtering and post-filtering methods. The MS welding chip is in the form of a cylindrical disc. This paper also considers the effect of lightening. The image is acquired from a progressive scan camera and Quartz Tungsten lighting is used. Thus by using all these techniques, the experimental results show that the proposed algorithm is suitable to the condition of the real time processing and accuracy of detection process in Machine Vision was found to be 96.31%.