A method for detecting and recognizing wallpaper defects based on OTSU and GA-BP neural network

Abstract

The invention discloses a method for detecting and recognizing wallpaper defects based on OTSU and GA-BP neural network, comprising the steps of acquiring a detection image of the wallpaper to be detected, and preprocessing the detection image to obtain a preprocess image by adopting an RGB color function; Calculating a proportion of pixel points whose pixel value is smaller than a pixel thresholdvalue in the preprocessed image; by using OTSU threshold segmentation method, performing defect segmentation of the preprocessed image when the occupation ratio is larger than a set threshold value;calculating the gray-scale and geometric features of the defects in the segmented image; Inputting the gray level feature and the geometric feature into a pre-trained GA- BP neural network to detect the defects in the wallpaper and obtaining the types of the defects; When the occupation ratio is less than or equal to a set threshold, determining that the wallpaper to be inspected is free of defects.

Description

technical field [0001] The invention relates to image detection technology, in particular to a wallpaper defect detection and recognition method based on OTSU and GA-BP neural network. Background technique [0002] Wallpaper, also known as wallpaper, is an interior decoration material used to decorate walls. In the actual production line, due to mechanical aging and other reasons, the produced wallpaper products contain defects. The most common defects are wrinkles, holes, cracks, and black spots. [0003] The current defect detection and recognition is mainly divided into two parts: defect segmentation and defect recognition and classification. In terms of defect segmentation, threshold segmentation is a common method for defect segmentation. Since the traditional threshold segmentation method can no longer meet the detection requirements of defect images, many studies have improved the traditional threshold segmentation method and used it for defect image segmentation det...

AI Technical Summary

Problems solved by technology

In terms of defect segmentation, threshold segmentation is a common method for defect segmentation. Since the traditional threshold segmentation method can no longer meet the detection requirements of defect images, there have been many studies to improve the traditional threshold segmentation method and use it for defect image segmentation detection.

Aiger D and Talbot H used the fast Fourier transform of the original image to obtain the spectrum image, and then used PHOT to filter out the background texture to obtain the defect. Although this method can accurately segment the defect, the defect area after segmentation is distorted

[0004] Seba Susa and others built a Gaussian mixture model to automatically detect defects, but it has a large error in defect detection in complex backgrounds

Use robust principal component analysis (Robust Principal Component Analysis, RPCA) to decompose the image, and obtain the defect image from the sparse matrix after binary decomposition. Although this method can segment the defect from part of the original image, it is only applicable to Defect extraction on a smooth and simple background, the background of the wallpaper has a great influence on the segmentation effect

Images