摘要
在颗粒加工工业,颗粒尺寸和形状参数的获取是一道常见的工序;体积是一个重要的颗粒三维参数,采用传统的手工测量方法获取体积耗时长,人工投入较多,很难实现过程控制中的实时反馈;应用计算机视觉技术,提出一种基于颗粒单视二维图像信息(周长、投影面积、长宽比等)的BP神经网络体积估算方法;实验结果表明,BP神经网络体积估算模型的非线性映射能力能够很好地反应多个影响参数和体积之间的复杂关系,具有较好的精确性、可行性、适应性。
In the particle processing industry, it is a common procedure to obtain the size and shape parameters of the particles. Obtai- ning the volume of the particles by means of the conventional approach is time--consuming and labor--intensive, which is impossible for real --time feedback in process control. Utilizing the computer vision techniques, this paper proposes an approach for the volume estimation of the particles by applying the BP neural network based on the 2--D image information of single view （perimeter. area. elongation etc）. The experimental result shows that the nonlinear reflection ability of BP neural network model can well describe the complex relationship between influencing factors and attain the volumetric estimation with sufficient precision, feasibility and adaptability.
出处
《计算机测量与控制》
CSCD
北大核心
2009年第3期571-572,575,共3页
Computer Measurement &Control
基金
宁波市自然基金项目(2006A610016)
国家教育部留学回国基金资助项目(2006699)
关键词
单视二维图像
特征参数
神经网络
体积估算
particle image
particle parameters
neural network
volume estimation
