船载移动视频电子稳像方法研究

发布时间：2018-05-21 08:54

本文选题：船载视频 + 电子稳像　；参考：《大连海事大学》2017年博士论文

【摘要】：船载移动视频技术能够弥补船舶自动识别系统和雷达在海上弱小目标探测方面的不足,提高船舶的情景感知能力。然而,船载视频系统受船舶六自由度运动和振动等因素的影响,会产生颠簸和摇摆导致图像抖动,不利于目标检测、跟踪和识别。因此,本文研究船载移动视频的电子稳像方法,滤除图像抖动,为目标检测、跟踪和识别等应用提供稳定的视频数据。本文首先分析了船载移动视频特征,其次实现了惯性数据和视频数据的同步采集,最后分别研究了基于特征点匹配和振动测量以及基于海天线检测和遗传算法的船载移动视频电子稳像算法。在船载移动视频特征分析方面,利用惯性传感器测量船舶和摄像机振动特征,并探究其内在关系,结合摄像机成像模型,分析了图像序列的运动机理,为电子稳像研究奠定了理论基础。在数据采集方面,利用惯性传感器控制摄像机曝光的方式,实现数据同步采集,为基于惯性数据的电子稳像方法的研究提供了试验基础。在数据同步的基础上,首先研究了基于特征点匹配和振动测量的电子稳像算法,提出了自适应Harris角点检测算法,利用K均值算法自适应选取目标周围强度大且稳定的角点,基于角点的运动特征,提出了基于傅里叶级数曲线拟合和基于振动测量的运动滤波算法,试验表明,图像序列的均方误差大约降低了 68%;其次,研究了基于海天线检测和遗传算法的电子稳像算法,提出了基于惯性测量和Hough变换融合的海天线检测算法,根据当前帧海天线位置和角度,结合帧间摄像机惯性数据,预测下一帧海天线区域,试验表明,海天线检测区域面积缩小了 80%-96%,检测准确率达到100%;此外,利用相邻帧海天线运动估算图像运动矢量,建立了图像低频晃动滤波模型,利用遗传算法对模型进行求解,该模型可实现在图像损失一定的情况下,稳像效果最佳,试验表明,稳像处理后图像序列的均方误差降低了 36%-62%。本文研究成果提高了船载移动视频的稳定性,提升船舶避碰、防海盗以及海上遇险船舶和落水人员的搜寻救助能力,对保障海上人命财产安全发挥重要作用。
[Abstract]:Shipborne mobile video technology can make up for the shortage of ship automatic recognition system and radar in detecting small and weak targets at sea and improve the capability of ship's scene perception. However, the shipborne video system is affected by the motion and vibration of the ship with six degrees of freedom, which will lead to the image jitter, which is not conducive to target detection, tracking and recognition. Therefore, this paper studies the electronic image stabilization method of shipborne mobile video, filters image jitter, and provides stable video data for target detection, tracking and recognition applications. In this paper, we first analyze the characteristics of ship moving video, and then realize the synchronous acquisition of inertial data and video data. Finally, the electronic image stabilization algorithms based on feature point matching and vibration measurement, sea antenna detection and genetic algorithm are studied respectively. In the aspect of ship moving video feature analysis, the vibration characteristics of ship and camera are measured by inertial sensor, and the intrinsic relationship is explored. The motion mechanism of image sequence is analyzed by combining the camera imaging model. It lays a theoretical foundation for the study of electronic image stabilization. In the aspect of data acquisition, the method of controlling camera exposure by inertial sensor is used to realize synchronous data acquisition, which provides the experimental basis for the research of electronic image stabilization method based on inertial data. On the basis of data synchronization, the electronic image stabilization algorithm based on feature point matching and vibration measurement is studied, and an adaptive Harris corner detection algorithm is proposed. The K-means algorithm is used to adaptively select the strong and stable corner around the target. Based on the motion characteristics of corner points, a motion filtering algorithm based on Fourier series curve fitting and vibration measurement is proposed. The experimental results show that the mean square error of image sequence is reduced by 68%. The electronic image stabilization algorithm based on sea antenna detection and genetic algorithm is studied. A sea antenna detection algorithm based on the fusion of inertial measurement and Hough transform is proposed. According to the position and angle of the current frame sea antenna, the camera inertia data between frames are combined. The prediction of the next frame area of sea antenna shows that the area of detection area of sea antenna is reduced by 80% -96 and the detection accuracy is 100%. In addition, the image motion vector is estimated by the motion of adjacent frame sea antenna, and the image low frequency sloshing filter model is established. The genetic algorithm is used to solve the model. The model can achieve the best image stabilization effect under the condition of certain image loss. The experiment shows that the mean square error of image sequence after image stabilization is reduced by 36-62. The research results of this paper improve the stability of shipborne mobile video, enhance the ship collision avoidance, anti-piracy, ships in distress and drowning personnel search and rescue ability, and play an important role in ensuring the safety of life and property at sea.
【学位授予单位】：大连海事大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：U665.2;TP391.41

【相似文献】