基于CLAHE与卡尔曼滤波的掘进机机载视频稳像算法  

An onboard video stabilization algorithm for roadheader based on CLAHE and Kalman filter

作　　者：李成城 马立森 田原 贾运红 贾曲 田伟琴 张凯 LI Chengcheng;MA Lisen;TIAN Yuan;JIA Yunhong;JIA Qu;TIAN Weiqin;ZHANG Kai(China Coal Research Institute,Beijing 100013,China;CCTEG Taiyuan Research Institute,Taiyuan 030006,China;China National Engineering Laboratory for Coal Mining Machinery,Taiyuan 030006,China;Shanxi Tiandi Coal Mining Machinery Co.,Ltd.,Taiyuan 030006,China)

机构地区：[1]煤炭科学研究总院,北京100013 [2]中国煤炭科工集团太原研究院有限公司,山西太原030006 [3]煤矿采掘机械装备国家工程实验室,山西太原030006 [4]山西天地煤机装备有限公司,山西太原030006

出　　处：《工矿自动化》2023年第5期66-73,共8页Journal Of Mine Automation

基　　金：国家重点研发计划项目(2020YFB1314002)。

摘　　要：掘进机等煤机装备在行进或作业期间,易因车体振动引起机载相机视频模糊,导致基于机载视频的机器视觉检测精度和可靠性下降。针对该问题,提出一种基于CLAHE与卡尔曼滤波的掘进机机载视频稳像算法。该算法由运动估计、轨迹平滑和运动补偿3个部分组成。在运动估计阶段,先采用限制对比度自适应直方图均衡(CLAHE)算法对井下巷道图像进行增强处理,再利用Shi-Tomasi算法获取每帧图像的特征点,对获取的特征点进行光流追踪和匹配,进而计算出相机的运动轨迹。在轨迹平滑阶段,利用卡尔曼滤波,根据视频前一帧的最优值预测当前时刻值,避免均值滤波需预先存储采样数据的问题,提高稳像的实时性。在运动补偿阶段,根据原始运动路径和平滑路径的关系对抖动视频逐帧补偿,生成稳定的视频序列。实验结果表明:(1)经CLAHE增强处理后,特征点匹配成功率比未增强处理时提高了58%,比HE增强处理时提高了43%,说明CLAHE算法可有效提高图像特征点匹配数。(2)通过像素偏移分析、差分图分析、峰值信噪比(PSNR)分析,验证了基于CLAHE与卡尔曼滤波的掘进机机载视频稳像算法具有较好的稳像效果。(3)与传统的HE+均值滤波算法相比,基于CLAHE与卡尔曼滤波的算法处理100帧视频图像的整体耗时减少了0.379 s,在去除抖动的同时,有效提高了稳像的实时性。During the movement or operation of coal mining equipment such as a roadheader,the vibration of the vehicle body can easily cause blurring of the onboard camera video.This leads to a decrease in the precision and reliability of machine vision detection based on the onboard video.In order to solve the above problem,an onboard video stabilization algorithm for roadheader based on CLAHE and Kalman filter is proposed.This algorithm consists of three parts:motion estimation,trajectory smoothing,and motion compensation.In the motion estimation stage,the contrast limited adaptive histogram equalization(CLAHE)algorithm is used to enhance the image of the underground roadway.The Shi-Tomasi algorithm is used to obtain the feature points of each image frame.The obtained feature points are tracked and matched by optical flow,and then the motion trajectory of the camera is calculated.In the trajectory smoothing stage,Kalman filtering is used to predict the current time value based on the optimal value of the previous frame of the video.It avoids the problem of prestoring sampling data of mean filtering and improves the real-time performance of image stabilization.In the motion compensation stage,the jitter video is compensated frame by frame based on the relationship between the original motion path and the smooth path,generating a stable video sequence.The experimental results show the following points:①After CLAHE enhancement processing,the success rate of feature point matching is increased by 58%compared to the non-enhancement processing and 43%compared to the HE enhancement processing.It indicates that the CLAHE algorithm can effectively improve the matching number of image feature points.②Through pixel offset analysis,differential image analysis,and peak signal-to-noise ratio(PSNR)analysis,it is verified that the onboard video stabilization algorithm for roadheader based on CLAHE and Kalman filter has a good image stabilization effect.③Compared with the traditional HE+mean filtering algorithm,the algorithm based o

关 键 词：智能掘进 掘进机 机载视频稳像 图像增强 限制对比度自适应直方图均衡 卡尔曼滤波 

分 类 号：TD632[矿业工程—矿山机电]