基于复杂网络时空特性的泡沫图像动态纹理特征分析

发布时间：2018-09-13 08:12

【摘要】：基于机器视觉的浮选过程监控方法已经被广泛应用于浮选过程中,泡沫表面纹理特征是过程监控的关键视觉特征之一。当前静态纹理特征只能从空间维度描述图像特征,在时间维度上刻画图像序列的内在变化特性存在不足,不能准确反映浮选泡沫浮选过程动态特性。提出了基于复杂网络时空特性的泡沫图像序列动态纹理特征方法。通过将每帧图像的像素点映射到网络各节点,利用邻接矩阵建立复杂网络模型和网络权值动态演化反应不同时刻的图像特征,基于复杂网络时空特性提取泡沫图像序列的动态纹理特征。结合实际生产数据进行仿真验证,实验结果表明该方法可准确识别浮选动态状况,为浮选生产过程的实时调节提供重要的指导信息。
[Abstract]:Flotation process monitoring method based on machine vision has been widely used in flotation process. The texture feature of foam surface is one of the key visual features of process monitoring. At present, the static texture features can only describe the image features from the spatial dimension, but the inherent variation characteristics of the image sequences can not be described in the time dimension, which can not accurately reflect the dynamic characteristics of flotation foam flotation process. A dynamic texture feature method for foam image sequences based on spatio-temporal characteristics of complex networks is proposed. By mapping the pixel points of each frame image to each node of the network, the complex network model and the dynamic evolution of network weights are established by using the adjacency matrix to reflect the image features at different times. The dynamic texture features of foam image sequences are extracted based on the temporal and spatial characteristics of complex networks. The experimental results show that this method can accurately identify the dynamic state of flotation and provide important guidance information for the real-time adjustment of flotation production process.
【作者单位】：中南大学信息科学与工程学院;
【基金】：国家自然科学基金项目(61473319);国家自然科学基金创新研究群体项目(61321003) 中南大学创新驱动计划项目(2016CX014)~~
【分类号】：TP391.41

【相似文献】