针对目标探测器在大气中高速飞行时受湍流干扰, 导致光学系统接收到的视频/图像产生像素偏移、模糊、信噪比降低等问题, 本文对湍流退化视频/图像复原的复杂性及复原方法进行了研究, 提出了一种基于非凸势函数优化与动态自适应滤波的湍流退化视频复原方法。首先, 研究了湍流退化视频的求和与去模糊框架, 并通过利用非刚性配准方法对刚性全局配准方法进行改进, 进一步缩小了模糊核的尺度; 然后, 在计算机视觉的非凸优化框架下, 构建了图像解卷积的非凸性算法, 有效地解决了图像解卷积难题; 最后, 结合湍流退化视频自身特点, 对超分辨率视频复原的动态自适应滤波框架进行了扩展与改进, 使其适用于湍流退化视频的复原。仿真实验结果表明, 本文方法的复原效果不仅有较大提升, 而且实现了对湍流退化视频序列的动态自适应复原。

The video/image received by optical system suffers degradation such as pixel deviation, blurring and signal to noise ratio reduction, when the target detector flies at high speed in the atmosphere is affected by turbulent interference. In this paper, the complexity of turbulence-degraded image restoration and deblurring methods are studied, and a video restoration method based on non-convexity optimization and dynamical adaptive filtering method is proposed. Firstly, we study the sum and deblur framework of turbulence-degraded video. By improving the rigid global registration with non-rigid local registration, the scale of blur kernel can be narrowed. Secondly, a non-convexity deconvolution algorithm is constructed based on the non-convexity optimization framework, which can solve the difficulty of image deconvolution problem effectively. Finally, dynamic adaptive filtering framework of video super-resolution recovery is expanded with the characteristics of turbulence-degraded. Simulation experiment results show that the proposed method can not only improve the quality of restored image, but also implement dynamic self-adaptive restoration of turbulence-degraded video.