Huaiguang Liu

Accurate and efficient white-spot defects detection for the surface of galvanized strip steel is one of the most important guarantees for the quality of steel production. It’s a fundamental but “hard” small target detection problem due to its small pixel occupation in low-contrast images. By fully exploiting the low-rank and sparse prior information of surface defect image, a Schatten-p norm-based low-rank tensor decomposition (SLRTD) method is proposed to decomposes the defect image into low-rank background, sparse defect, and random noise. Firstly, the original defect images are transformed into a new patch-based tensor mode through data reconstruction for mining valuable information of defect image. Then, considering the over-shrinkage problem in low-rank component estimation caused by vanilla nuclear norm and weighted nuclear norm, a nonlinear reweighting strategy based on Schatten p-norm is incorporated to improve the decomposition performance. Finally, a solution framework is proposed via a well-designed alternating direction method of multipliers to obtain the white-spot defect target image by a simple segmenting algorithm. The white-spot defect dataset from real-world galvanized strip steel production line is constructed, and the experimental results demonstrate that the proposed SLRTD method outperforms existing state-of-the-art methods qualitatively and quantitatively.

During the production process of steel plate, surface defect detection is crucial for high-quality products. For the existing defect detection method based on machine vision, there are various types of problems, such as large model calculations, low detection accuracy and difficulties of recognizing small defect targets. To reduce and solve these issues, the paper proposes a new defect detection model, simplified kernel and squeeze on a you only look once network (SKS-YOLO), which can achieve rapid and effective defect detection on steel plate. Firstly, it adopts EfficientNetv2 as the backbone, significantly reducing model calculations and accelerating training speed while maintaining accuracy. Subsequently, the atrous spatial pyramid pooling (ASPP) module is utilized to obtain a larger receptive field, extracting more feature information from surface defects. The integration of the squeeze excitation network (SE-Net) attention mechanism enhances capabilities of feature extraction furtherly. Then, the K-means algorithm is applied to cluster and obtain more suitable anchor frames for defect targets. It not only increases the number of positive samples, but also expedites model convergence. Finally, the loss function of simplified intersection over union (SIoU) is used to enhance the ability of model to locate and detect surface defect targets. The experimental results show that the mean average precision (mAP) is 89.40% at a detection speed of 55 frames per second (FPS), which is better than the state-of-the-art (SOTA) detection models.

The existence of defects in solar cells due to the process or material reasons in the production process. Based on the photoluminescence imaging principle, an image enhancement method for solar cells based on background assessment and a defect recognition method based on morphological feature and HOG feature fusion were proposed. Firstly, the characteristics of shape and location of cell defects were analyzed, and the two-step segmentation method was proposed to extract multi-directional HOG features from the segmented defects, and Laplace feature mapping method was adopted to reduce the dimension of HOG features. Then, the morphological characteristics such as aspect ratio and circularity were fused. Finally, according to the kernel function and penalty factor in support vector machines (SVM), the particle swarm optimization (PSO) algorithm was optimized to improve the defect classification effect. Fifty images were detected by using the proposed method, and the accuracy of classification recognition reached 98.3%. Comparing the proposed algorithm with the traditional SVM algorithm and Le-Net network, it can be seen that the proposed algorithm has the higher recognition accuracy.

The defects in photovoltaic cells affect the service life and power generation efficiency of the entire photovoltaic system. Aiming at the high missed detection rate of weak and small defects in the automatic detection of existing cells, a feature-enhanced lightweight convolutional neural network model was established. The feature enhancement extraction module was designed specifically to improve the extraction ability of weak boundaries. In addition, according to the principle of multi-scale recognition, a small target prediction layer was added to realize multi-scale feature prediction. In the experimental test, the mean average precision (mAP) of the model reaches to 87.55%, which is 6.78 percentage points higher than the traditional model. Moreover, the detection speed reaches to 40 fps, which meets the accuracy and real-time detection requirements.

The defects of solar cell component (SCC) will affect the service life and power generation efficiency. In this paper, the defect images of SCC were taken by the photoluminescence (PL) method and processed by an advanced lightweight convolutional neural network (CNN). Firstly, in order to solve the high pixel SCC image detection, each silicon wafer image was segmented based on local difference extremum of edge projection (LDEEP). Secondly, in order to detect the defects with small size or weak edges in the silicon wafer, an improved lightweight CNN model with deep backbone feature extraction network structure was proposed, as the enhancing feature fusion layer and the three-scale feature prediction layer; the model provided more feature detail. The final experimental results showed that the improved model achieves a good balance between the detection accuracy and detection speed, with the mean average precision (mAP) reaching 87.55%, which was 6.78% higher than the original algorithm. Moreover, the detection speed reached 40 frames per second (fps), which meets requirements of precision and real-time detection. The detection method can better complete the defect detection task of SCC, which lays the foundation for automatic detection of SCC defects.

The microstructure is the key factor for quality discriminate of coke. In view of the characteristics of coke optical tissue (COT), a segmentation method of coke microstructures based on adaptive clustering was proposed. According to the strategy of multiresolution, adaptive threshold binarization and morphological filtering were carried out on COT images with lower resolution. The contour of the COT body was detected through the relationship checking between contours in the binary image, and hence, COT pixels were picked out to cluster for tissue segmentation. In order to get the optimum segmentation for each tissue, an advanced -means method with adaptive clustering centers was provided according to the Calinski-Harabasz score. Meanwhile, Euclidean distance was substituted with Mahalanobis distance between each pixel in HSV space to improve the accuracy. The experimental results show that compared with the traditional -means algorithm, FCM algorithm, and Meanshift algorithm, the adaptive clustering algorithm proposed in this paper is more accurate in the segmentation of various tissue components in COT images, and the accuracy of tissue segmentation reaches 94.3500%.

Binocular vision is a popular method for 3d measurement, but the precision and efficiency is still a limitation. In order to solve the problem of low precision and high mismatching rate of current stereo match methods in weak texture region, a local stereo matching algorithm with multi‐scale based on anisotropic match cost was proposed. Not only including the gray value between the adjacent pixels as the absolute difference (AD) method, the matching cost function also involved gradient and phase information to make an anisotropic evaluation and eliminate the outliers of the flat area. Meanwhile, a multi‐scale strategy with the image pyramid was referenced, and the dynamic disparity search range of variable window was used on the original cost aggregation framework by improving the stereo matching algorithm of cross‐scale cost aggregation. The matching cost volume in each scale space used the dynamic support window to guide the filter to aggregate the matching cost. Finally, to overcome the problems of disparity selection ambiguity of WTA (winner‐take‐all) strategy and the horizontal fringe introduced by left‐right consistency (LRC) detection, the weighted median filter based on guided filter weight was used to carry out disparity refinement. According to the experiments, the method got more high matching accuracy, and the average error on the Middlebury test platform reached 5.25%.