一种基于时空相关性的星图降噪算法

摘要

This paper proposes a new method for denoising of star map based on spatio-temporal correlationship,which can solve the problem of low PSNR and star-point extraction difficulty in space exploration.Firstly,the energy characteristics of real star map and background noise are analyzed and modeled.Secondly,a feasible denoising algorithm is proposed based on the model.The algorithm is based on the spatio-temporal correlationship detection,and uses the detection result to locate the target and noise continuously.And then the target is matched and smoothed by the result of localization,and the background noise is suppressed,so as to improve the PSNR of the whole image.Finally,the validity of the algorithm is verified by simulation platform.The results show that the algorithm can improve the PSNR of the whole image effectively,especially the PSNR of point targets.The PSNR is increased by 1.6dB under certain noise background,and the histogram retain the goal pixels perfectly.The proposed algorithm is superior to most of existing noise reduction algorithms in similar computational complexity and is very likely to be applied to inorbit space exploration systems and has broad application prospects.%文章提出了一种基于时空相关性的星图降噪方法,用于解决空间探测时星图峰值信噪比(Peak Signal to Noise Ratio,PSNR)低、星点提取难度大的问题.文章首先对真实星图中星点和背景噪声的能量特性进行了分析与建模.然后,基于该模型提出了一种可行的降噪算法,该算法以时空相关性检测为核心,并利用检测的结果对目标和噪声进行连续多帧定位.进一步利用定位的结果对目标进行匹配和平滑,对背景噪声进行抑制,从而在总体上提高图像的PSNR.最后,基于仿真平台对算法的有效性进行了测试验证.结果表明,该算法能够有效提升星图PSNR,尤其是点目标的PSNR,在一定噪声等级下,可提升1.6dB.降噪之后,在去除噪声的同时,星图的灰度直方图完好保留了高灰度值部分的目标点像素.该算法在相似的运算复杂度下,其降噪效果明显优于绝大多数的现有降噪算法,利于后续提取星图中恒星及点目标,应用于在轨空间探测系统中,具有广阔的应用前景.