基于图像特征的嵌入式大包下渣检测方法  

Embedded Ladle Slag Detection Method Based on Image Characteristic

作　　者：李培玉[1] 郝明金[1] 王团结[1] 

机构地区：[1]浙江大学机械工程学系,浙江杭州310027

出　　处：《钢铁研究学报》2011年第1期58-62,共5页Journal of Iron and Steel Research

摘　　要：通过对现有大包下渣检测系统的分析,提出了一种基于图像特征的嵌入式下渣检测方法,阐述了系统嵌入式架构及具体实现方法。该方法以DSP为下位机系统的核心硬件平台,由FPGA与DSP组成的下位机系统将摄像头采集到的模拟图像转换成数字图像,同时进行平均灰度变化与结构相似度的图像特征提取与分析,再经HPI通信模块将处理后的数据发送给由ARM组成的上位机系统,最后采用MiniGUI设计了嵌入式环境下的图形用户界面,实现对大包浇注过程中长水口附近液面状态及下渣检测情况的实时监控。现场试验表明:该方法具有非接触式检测、成本低等特点,能够协助现场工作人员检测和判断下渣,有效地减少连铸过程中大包到中间包的下渣量。An embedded slag detection method in which an image characteristic technology was used to detect the slag was proposed by analyzing the existing bale slag detection systems. The system embedded architecture and detail implementation processes were described. Based on lower position computer system composed with FPGA and DSP which was core hardware platform, analog image sampled by camera was converted into digital image. Changes in average gray and structural similarity were extracted and analyzed. Then the processed image data was transmitted to upper position computer system composed with ARM by the HPI communication module. At last, the graphic user interface based on MiniGUI was designed in embedded environment to realize the real--time stream monitoring in the liquid state of tundish and the slag detection progress. Field experimental results show that the stag detecting system design scheme has following characteristics： non-contact detection, low cost and so on. It can help field workers to judge whether slag carry-over occurs or not, in order to effectively minimize slag amount leaked into tundish from bale.

关 键 词：图像特征 嵌入式系统 下渣检测 图像处理 

分 类 号：TF777[冶金工程—钢铁冶金]