冶金焦炭图像的拍摄与自动分析技术  

Imaging and Automated Analysis of Metallurgical Coke

作　　者：谢威[1] 夏杰[1] 钱晖[2] 陈雁秋[1] 

机构地区：[1]复旦大学,计算机科学技术学院 [2]宝钢集团有限公司中央研究院

出　　处：《微型电脑应用》2015年第10期14-17,4,共4页Microcomputer Applications

基　　金：国家自然科学基金面上项目(61175036)资助

摘　　要：冶金焦炭光学组织的测定对于评价焦炭质量十分重要。但是现有的测量技术,还局限在人工数点计数法。为了自动化的高精度测定焦炭光学组织,提出了一种新的冶金焦炭光学组织测定方法。该方法首先构建了一套自动化拍摄冶金焦炭光学组织双反射率图片的硬件系统,然后通过验证摄像机的线性区域,并使用统一的像素值到反射率的转化关系,提高了系统的拍摄的准确性和精度。最后,为了自动地识别,提出了使用反射率图片的纹理信息来分割识别不同的光学组织成分的算法。实验证明该系统拍摄精度可靠,分割识别的准确率也超过了国际上主流的算法。Measurements of optical texture of metallurgical cokes are important in coke quality assessment.To date,these measurements have been limited to manually point-counting measurements.In this work,a new method is presented to measure optical texture of coke.The system which can automatically acquire bi-reflectance images of coke is built.The accuracy of this system is improved by finding the linear range of camera and using a uniform linear correlation to convert pixel values into reflectance.It also proposes a novel algorithm which takes advantage of texture information of reflectance images to recognize different optical textures.This segmentation algorithm outperforms other state-of-art algorithms.

关 键 词：冶金焦炭 双反射率图像 摄像机标定 GABOR特征 图像分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]