矿井车载摄像系统中电子稳像算法的研究

为克服矿井车载摄像系统在视频采集过程中的图像抖动问题,提出了一种基于MIC-Harris角点特征的自适应运动补偿电子稳像算法。首先,提出MIC-Harris角点检测算法和由粗到精的匹配策略,实现序列中特征点的快速精确提取。然后根据矿井移动车载图像运动特点,构建运动参数估计模型,运用最小二乘法实现全局运动矢量的求解。最后,提出自适应相邻帧补偿方法,结合Kalman滤波实现补偿矢量的计算,在滤除高频随机抖动同时保留了车载摄像机低频主动运动,并克服了连续视频序列长时间稳像处理的帧间累积误差问题。实验仿真结果表明：本文提出的算法不仅融合了MIC运算速度快和Harris准确度高的优点,而且能够实现矿井车载摄像系统的长时间连续稳像处理,具有较好的稳像效果。

An electronic image stabilization of camera system for underground vehicle carries

An adaptive motionScompensation electronic image stabilization based on MIC-Harris corner feature was proposed to solve the jitter problem in the video acquired from underground-vehicle camera system. Firstly, the MIC-Harris corner detection algorithm and the matching strategy from coarse to precious are presented to extract feature points accurately in the sequence. Then, according to the motion characteristics of moving vehicle image in mine, the motion parameter estimation model is build, and the global motion vectors are computed by the least squares method. Finally, combined with calculation of compensation vectors by Kalman filter, the adaptive compensation method of adjacent frames is proposed. The high frequency random jitter is filtered while the low frequency active motion is retained, and the inter cumulative error problem of prolonged stabilization process issolved. The experiments show that the proposed algorithm in this paper can not only fuse the fast computing speed of MIC and high accuracy of Harris, but also achieve long continuous stabilization processing of underground-vehicle camera system. The proposed algorithm has a good stabilization effect.