Foreground segmentation and location of coal and gangue under complex similar background

Abstract

To improve the foreground segmentation and location accuracy of complex coal gangue images with gray histogram distribution close to the unimodal shape, a contour detection algorithm of the grayscale fluctuation matrix is proposed. The contour and non-contour pixels of coal and gangue images are investigated, and the result indicates that the gray values of the pixels around the contour exhibit the non-uniform distribution, and the gray value changes in different directions are significantly different. Accordingly, a grayscale fluctuation matrix is built by calculating the change amplitude of pixels in different directions, and multiple features are extracted from the grayscale fluctuation matrix to realize the target contour segmentation. Furthermore, the contour is optimized using the historical and future information of the contour image, thus effectively removing numerous false contours, reproducing some hidden contours and increasing segmentation accuracy. Compared with the artificial segmentation results, the pixel area and centroid coordinates of the proposed method achieve the maximum error rates of 4.404% and 3.18%, respectively, all lower than 4.5%. This study provides a feasible solution to the edge detection and segmentation of images with similar and complex backgrounds.

A coal and gangue detection method for low light and dusty environments

“…The model's accuracy and robustness were validated under different lighting intensities. Luo et al [18] introduced a gray-level co-occurrence matrix contour detection algorithm, providing a viable solution for edge detection and segmentation of similar and complex background images.…”

Multiple damage segmentation and extraction of wind turbine blades surface under complex background based on SKRT approach

“…Complex background often has a significant impact on blade damage detection [26]. In order to reduce the interference of complex background on blade damage detection, relevant studies found that the ratio of the maximum and minimum values of contour pixels and non-contour pixels showed strong separability by constructing a grayscale fluctuation matrix, and on this basis, low-contrast features of complex similar background and targets were extracted [27]. For low contrast damage and background characteristics of blade surface with complex background caused by non-uniform illumination, the illumination model of cartoon mapping was established based on Gaussian scale space to remove nonuniform illumination, then enhance surface damage and image background contrast through multi-direction Gabor transform [28].…”

Research on coal gangue recognition method based on XBS-YOLOv5s

“…At present, the image method [10][11][12] is more widely used in the field of gangue sorting because of its advantages of noncontact identification, high efficiency, accuracy and diversity. Among them, the image method based on deep learning [13][14][15] has the advantages of no need for feature selection, strong feature expression, fast detection speed, and high detection accuracy compared to machine learning [16][17][18], thus gradually becoming a research hotspot. Yan et al [19] used random forest algorithm and band correlation analysis to select the best spectral image of coal gangue, and introduced the GIoU loss function and DIoU-NMS function in YOLOv5s algorithm to improve the detection accuracy of the model for overlapping targets and small targets.…”