摘要
在铁矿石烧结过程中,烧结机尾断面图像中的火焰区域蕴含着大量烧结特征信息。为有效识别烧结断面火焰核心区域,提出了改进的粒子滤波算法对烧结断面图像进行火焰区域跟踪。在经典粒子滤波算法中引入RGB颜色空间生成带权粒子。同时,烧结过程中火焰区域可能合并、分裂及脱落,对粒子滤波算法中重采样方法进行权值优化、粒子舍弃与重采样策略优化,避免由粒子退化现象与样本贫乏问题引起跟踪失败。结果表明,改进的粒子滤波算法能够对烧结断面图像火焰区域进行有效、快速地跟踪。
In the iron ore sintering process,the flame area in the cross-sectional image of the tail of the sintering machine contains a lot of sintering characteristic information.In order to effectively identify the flame core area of the sintering section,this paper proposes an improved particle filter algorithm to track the flame area of the sintering section image.In the classic particle filter algorithm,RGB color space is introduced to generate weighted particles.At the same time,the flame area may merge,split and fall off during the sintering process.The weight optimization,particle rejection and resampling strategy optimization of the re-sampling method in the particle filter algorithm are carried out to avoid tracking failures caused by particle degradation and sample shortage problems.The results show that the improved particle filter algorithm can effectively and quickly track the flame area of the sintering section image.
作者
王福斌
何江红
武晨
WANG Fubin;HE Jianghong;WU Chen(College of Electrical Engineering,North China University of Science and Technology,Tangshan Hebei 063210,China;Tang Steel International Engineering Technology Corp,Tangshan Hebei 063000,China)
出处
《激光杂志》
CAS
北大核心
2021年第12期94-101,共8页
Laser Journal
基金
河北省自然科学基金(No.F2019209323)。
