Machine vision has been widely used in industrial component production and processing due to its high accuracy, high efficiency and non-contact measurement. The electronic water pump of vehicle includes a gasket,which is sleeved on the rotor between pump body and the cylinder head joint surface and has the characteristics of stamping, sealing, high temperature resistance and so on. Therefore, in the process of pump production, gasket detection is very important link. The traditional manual visual inspection has some shortcomings, such as strong subjective factors, and the component is vulnerable to secondary damage. On this basis, this paper design a ceramic gasket state detection system based on machine vision, including software and hardware platforms, which can detect the state of the current components in real time and feed back to the inspection workers,reducing the subjectivity brought by manual operation, reducing the probability of components damage and improving the production efficiency.