Industrial product surface defect detection method based on FCN+FC-WXGBoost

Abstract

The invention discloses an industrial product surface defect detection method based on FCN+FC-WXGBoost. The method comprises the following steps: 1) performing size standardization and normalization operation on a picture; 2) carrying out online enhancement and batching on the picture; 3) sending the picture to a network combining a full convolution network (FCN) and a full connection network (FC)for training; 4) taking the input of an output layer of the full connection network as a feature vector, training a WXGBoost classification model, using a clone selection algorithm to carry out automatic parameter adjustment, replacing an output layer of a full connection layer with WXGBoost, and combining with a full convolution network FCN to obtain an FCN+FC-WXGBoost network model; 5) during detection, inputting the picture into the FCN+FC-WXGBoost network to obtain the position and category information of the defect. According to the method, the influence of conditions such as illumination, exposure and displacement on defect detection is effectively reduced, the defect detection stability is improved, the influence of defect category imbalance on the detection precision is reduced, and the detection precision is improved.

Description

technical field [0001] The invention relates to the technical field of surface defect detection of industrial products, in particular to a method for detecting surface defects of industrial products based on FCN+FC-WXGBoost. Background technique [0002] Defect detection is an important part of the production process, which guarantees the reliability of industrial products. The surface defect detection of industrial products requires precise positioning of the position of the defect and classification of the located defects. Prior to this, the surface defect detection technology of industrial products usually used traditional machine vision technology to perform operations such as image grayscale binarization, edge contour extraction, and template matching. The disadvantage of this type is that it is very sensitive to changes in image illumination and displacement. , with poor robustness. At the same time, the number of occurrences of different types of defects is indeterm...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to overcome the problems that the existing methods have high requirements on the conditions such as illumination changes and displacements of industrial product surface pictures, poor detection stability, and unbalanced defect sample categories, etc., and propose a method based on FCN+FC -WXGBoost's surface defect detection method for industrial products can effectively reduce the influence of light, exposure and displacement on defect detection, improve the stability of defect detection, and reduce the impact of defect category imbalance on detection accuracy, improving detection accuracy , which has the advantages of end-to-end, strong generalization, and high precision in defect detection

Images