GB/T 46608-2025Non-destructive testing of steel tubes - General rules for the machine vision inspection method for surface defects (English PDF)

3.1 Machine visual inspection

Machine vision converts the captured information about the steel pipe surface into image signals, which are then transmitted to a dedicated image processing system for analysis based on pixels.

Information such as brightness and color is converted into digital signals. The image system performs various operations on these signals to extract the target's features.

A detection method that uses target features to determine whether there are defects on the outer surface of the steel pipe being tested.

Note. The detection is carried out using two-dimensional and/or three-dimensional detection methods.

3.2 Data Acquisition Server

The device responsible for interfacing with the camera to acquire images.

Note. The design needs to be tailored to the testing speed and sampling frequency of the steel pipe.

3.3 Inference server

A device for detecting and classifying defects in real-time images of the steel pipe surface.

Note. The server performance meets the requirements for real-time performance, data querying, storage, and other aspects.

3.4 User terminal

The software provides users with various surface quality features, including defect alarms, defect distribution previews, and historical data queries, as well as a systematic approach.

A device that performs functions such as system maintenance and defect sample collection.

3.5 Line structured light three-dimensional depth detection

3D reconstruction is performed based on the principle of triangulation. During image acquisition, the line laser beam projects onto a plane of light; for... […]