Surface defect detection method and device and electronic device

Abstract

The embodiment of the invention provides a surface defect detection method and device and an electronic device, and the method comprises the steps: obtaining a to-be-detected image; inputting an image to be detected into a deep learning network model obtained through pre-training, obtaining a defect-free reconstructed image corresponding to the image to be detected, and the deep learning networkmodel being a deep learning network model obtained through pre-training based on a sample training set containing multiple defect-free training sample images; subtracting the reconstructed image fromthe to-be-detected image to obtain a difference image; and in the difference image, determining that a surface defect exists in a corresponding region in the to-be-detected image if a region of whichthe difference is greater than a preset difference exists. Through the scheme, the detection rate of the surface defects can be improved.

Description

technical field [0001] The invention relates to the technical field of target detection, in particular to a surface defect detection method, device and electronic equipment. Background technique [0002] Surface defect detection is to judge whether there are spots, pits, color differences, scratches, defects and other defects on the surface of the product. The surface defects of the product have the characteristics of various types, changeable shapes, unstable positions and diverse background textures. In the industrial field, surface defects directly affect the product's aesthetics, performance and other attributes. Therefore, the surface quality of the product is very important. The traditional surface defect detection is to equip a quality inspector on the production line, and the quality inspector conducts quality inspection on the product surface. Inspection, to judge whether there are defects on the surface of the product through human eye observation. However, this m...

AI Technical Summary

Problems solved by technology

However, machine learning requires that the sample image must contain a defective sample image, and the position of the defect on the product needs to be calibrated manually. However, due to the limited number of sample images containing defects, it cannot guarantee coverage of possible problems. All types of defects, therefore, the above methods cannot detect defects that are not present in the sample image

Images