【摘要】 当前煤泥浮选泡沫分类研究多针对光照充足条件下泡沫图像,对于夜晚车间光照不足的暗淡图像效果不好。针对这一问题,引入一种利用深度学习的有效浮选泡沫分类方法,建立了一个深度卷积神经网络同时执行特征学习与泡沫分类,逐层运算抽取图像本质信息,过滤光线影响。实验结果表明,在白天强光和夜晚弱光下,无需图像增强等预处理均获得很高的准确率,实现浮选泡沫端到端分类,提高了识别的抗干扰能力。更多还原

【Abstract】 The current studies about the froth of coal floatation are mainly aimed at the froth image under adequate light conditions, which can′t obtain good result from the gray image at night. In order to solve this problem, introduces an efficient flotation froth classification method with deep learning.Construct a deep convolutional neural network to perform feature learning and froth classification simultaneously, and it operates layer by layer to extract the essential information of image, filtering the impact of light. The experimental results show that the network has high accuracy in the daytime strong light and low light at night without image enhancement or other preprocessing. And the end-to-end classification of the flotation froth is achieved. The network improves the anti-jamming ability.更多还原