Review of optical detection technologies for inner-wall surface defects

Inner-wall structures are common, and as their carrier, inner-wall-shaped parts are widely used in various fields of industry and scientific research. The inner-wall structure is usually used to transport substances such as fluid and gas, or to interact mechanically with cylindrical structures. No matter what kind of use, there is no doubt that the defects on inner-wall surfaces will directly affect their application performance, so it is necessary to detect the defects of the manufactured inner-wall-shaped parts to ensure their surface quality. The defect detection technologies based on optical principles have the obvious advantages of non-destructive, high efficiency and flexibility, and have no strict requirements on materials, so it is a popular inner-wall detection way. Because of the particularity of inner-wall structures, after obtaining the inner-wall surface information with high signal-noise ratio through the traditional qualitative detection technologies or three-dimensional quantitative detection technologies, it is necessary to design corresponding algorithms to identify and extract defects. With the development of machine learning research, defect recognition algorithms are not limited to the traditional algorithms based on defect feature design logic judgment, but also derive a new idea to identify the existence and types of defects based on machine learning training model. The accuracy of defect recognition is effectively improved. This paper analyzes and summarizes the current mainstream technologies of defect recognition based on optical principles, and summarizes the principles, key technical points and research status of various methods. According to the development characteristics of the demand for inner-wall-shaped parts, it is inferred that the development trend of inner-wall defect detection is mainly around the three directions of making the detection system on-line, quantitative and efficient.