Overview of machine vision surface defect detection

The surface defects of industrial products have adverse effects on the aesthetics, comfort and performance of products, so the manufacturers detect the surface defects of products in order to detect and control them in time.

Machine vision detection methods can largely overcome the disadvantages of manual detection methods such as low sampling rate, low accuracy, poor real-time performance, low efficiency, and high labor intensity, and have been increasingly widely studied and applied in modern industry.

Taking machine vision surface defect detection as the research object, this paper reviews the application of machine vision in the field of surface defect detection on the basis of extensive investigation of related literature and development achievements. Analyzed the typical machine vision surface defect detection system working principle and basic structure, this paper expounds the current research status of surface defect visual inspection, the existing visual software and hardware platform, machine vision detection are reviewed in this paper involved image preprocessing algorithm, image segmentation algorithm, image recognition, image feature extraction and selection algorithm and other related theory and algorithm research, The basic ideas, characteristics and limitations of each method are summarized, and the possible development direction in the future is prospected.

In machine vision surface defect detection system, image processing and analysis algorithm is an important content, each algorithm has advantages and disadvantages and its scope of adaptation. How to improve the accuracy, real-time performance and robustness of the algorithm is always the direction of researchers' efforts.

Machine vision is the simulation of human vision, machine vision surface detection involves many disciplines and theories, how to make the detection further to the direction of automation and intelligent development, still need more in-depth research.

Surface defect detection

Machine vision technology is a non-contact, non-damage automatic detection technology, is an effective means to achieve equipment automation, intelligence and precision control, with safe and reliable, wide spectrum response range, can work in harsh environments for a long time and high production efficiency outstanding advantages. Machine vision detection system through the appropriate light source and image sensor (CCD camera) to obtain the surface image of the product, using the corresponding image processing algorithm to extract the feature information of the image, and then according to the feature information of the surface defect positioning, recognition, classification discrimination and statistics, storage, query and other operations;

The basic components of machine vision surface defect detection system

It mainly includes image acquisition module, image processing module, image analysis module, data management and man-machine interface module.

The image acquisition module is composed of industrial camera, optical lens, light source and clamping device, and its function is to complete the product surface image acquisition. Under the illumination of the light source, the product surface is imaged on the camera sensor through an optical lens, and the optical signal is first converted into an electrical signal, and then into a digital signal that can be processed by the computer. At present, industrial cameras are mainly based on CCD or CMOS chip cameras. CCD is the most commonly used image sensor in machine vision.

The machine vision light source directly affects the quality of the image. Its role is to overcome the ambient light interference, ensure the stability of the image, and obtain the image with as high contrast as possible. At present, halogen lamps, fluorescent lamps and light emitting diodes (LED) are commonly used. LED light source has been widely used for its small size, low power consumption, fast response speed, good luminescence monochromity, high reliability, light uniformity and stability, easy integration and other advantages.

According to the illumination method, the illumination system composed of light source can be divided into bright field illumination and dark field illumination, structured light illumination and strobe illumination. Bright field and dark field mainly describe the position relationship between the camera and the light source. Bright field illumination means that the camera directly receives the reflected light from the light source on the target. Generally, the camera and the light source are distributed on different sides, which is easy to install. Dark field illumination means that the camera indirectly receives the scattered light from the light source on the target. Generally, the camera is distributed on the same side as the light source, and its advantage is that it can obtain high contrast images. Structured light illumination is to project the grating or line light source onto the measured object, and demodulate the 3D information of the measured object according to the distortion generated by them. Strobe lighting is to illuminate the object with high frequency light pulse, and the camera shooting requires synchronization with the light source.

Image processing module mainly involves image denoising, image enhancement and restoration, defect detection and object segmentation. Because of the field environment, CCD image photoelectric conversion, transmission circuit and electronic components will make the image noise, these noise reduces the quality of the image and the image processing and analysis of the adverse impact, so the image should be preprocessed to denoise. Image enhancement is for a given image applications, purposefully stressed global or local characteristics of image, the original is not clear image clarity or emphasize certain features of interest, enlarge the differences between different object in the image characteristics, the characteristics of the inhibition was not interested in, to improve image quality, abundant information, Image processing methods to enhance the effect of image interpretation and recognition. Image restoration is a process of reconstruction or restoration of degraded images through computer processing. Image restoration often uses the same method as image enhancement, but the results of image enhancement still need to be verified in the next stage. Image restoration attempts to use the prior knowledge of the degradation process to restore the original appearance of the degraded image, such as the elimination of additive noise, motion blur restoration and so on. The purpose of image segmentation is to segment the target area in the image for further processing.

Surface defect detection applications

The application field is very wide, such as: iron and steel metallurgy, non-ferrous metal processing, high precision copper plate belt, aluminum plate belt, aluminum foil, stainless steel manufacturing, electronic materials, non-woven fabric, fabric, glass, paper, film.

The market for surface testing is very huge. For example, take a steel company with annual output of 10 million tons as an example. The market for the available surface testing equipment of this company is about 150 ~ 200 million yuan.

Why use a surface defect detection system? Ensure product quality, improve production process and reduce labor cost

Line scanning surface defect detection system mainly consists of:

The visual acquisition department mainly includes linear array camera, lens, light source and image acquisition card.

The system bracket includes: camera bracket, light source bracket, and console bracket.

The electrical part (communication/control part) includes encoders, motion control cards or PLCS, and possibly motors.

Other: all kinds of wires and cables, CL lines, power cords, SMPS, lighting controllers, etc.

There is a PC.