基于图像增强及改进霍夫检测算法的球团粒度检测  

Detecting of pellet size based on image enhancement and improved Hough detection algorithm

作　　者：张学锋 秦继洋 黄刘松 龙红明 余正伟 ZHANG Xuefeng;QIN Jiyang;HUANG Liusong;LONG Hongming;YU Zhengwei(Anhui Engineering Research Center for Intelligent Application and Security of Industrial Internet,Maanshan 243032,Anhui,China;School of Computer Information Engineering,Maanshan Normal College,Maanshan 243041,Anhui,China;School of Metallurgical Engineering,Anhui University of Technology,Maanshan 243032,Anhui,China)

机构地区：[1]安徽省工业互联网智能应用与安全工程研究中心,安徽马鞍山243032 [2]马鞍山师范高等专科学校计算机与信息工程学院,安徽马鞍山243041 [3]安徽工业大学冶金工程学院,安徽马鞍山243032

出　　处：《中国冶金》2024年第8期59-68,共10页China Metallurgy

基　　金：安徽省教育厅重点实验室项目(TZJQR007-2023);安徽省重点实验室开放基金资助项目(SKF24-04)。

摘　　要：为解决球团图像过曝失真及球团粘连难以识别的问题,提出一种改进的对比度限制自适应直方图均衡化方法实现图像增强,并引入多尺度霍夫检测算法实现对球团的有效检测。首先,引入细节保护机制,使用对比度增强函数来调整对比度限制自适应直方图均衡化(CLAHE)的输出结果,从而限制对比度增强的幅度,同时保留图像中更多的细节信息;其次,通过自适应中值滤波对增强图像噪声进行过滤;最后,引入多尺度的霍夫检测算法,解决粘连球团识别问题,实现对球团粒度的准确、实时、高效检测和统计。某钢厂实际应用结果表明,此方法对球团粒度识别的准确度达到94.44%,对合格球团个数的检测准确度可达96.72%,能够达到球团制备工艺的要求,对提高生产效率、降低人工成本、提高球团制备合格率等具有积极作用。To address the issues of overexposure distortion and challenging identification of pellet adhesion in pellet images,an enhanced contrast-limited adaptive histogram equalization method was proposed for image enhancement,along with the introduction of a multi-scale Hough detection algorithm for efficient pellet detection.Firstly,the en-hanced contrast-limited adaptive histogram equalization method incorporates a detail security mechanism by utilizing a contrast enhancement function to adjust the output of contrast-limited adaptive histogram equalization(CLAHE).The extent of contrast improvement is limited by this approach,while simultaneously preserving more detailed infor-mation in the images.Then,the image noise induced after enhancement is filtered by the adaptive median filtering method.Finally,the multi-scale Hough detection algorithm is applied to tackle both adhesion pellet identification and detection challenges,so as to realize the accurate,real-time and efficient detection and statistics of pellet size.Practical application results from a steel factory reveal that this method achieves a precision rate of as high as 94.44%,with 96.72%accuracy in determining the number of qualified pellets,fulfilling the requirements of pellet preparation process.This method plays a vital role in enhancing production effectiveness,decreasing labor expen-ses,and enhancing overall quality control rates for pellet preparation.

关 键 词：粒度检测 氧化球团 图像增强 霍夫检测 直方图均衡化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TF046.6[自动化与计算机技术—计算机科学与技术]