基于能量函数的极值中值滤波星图去噪算法

发布时间：2018-12-25 17:01

【摘要】：利用地基观测相机拍摄的以深空为背景的星图受星空复杂背景的影响,往往具有较高的噪声水平。同时由于星图主要由恒星、空间目标和星空背景噪声组成,且成点状分布,星图目标和噪声呈现较大的相似性,传统的图像去噪算法并不适用于星图。为此,该文提出一种基于能量函数的极值中值滤波去噪算法,该算法在去除星图椒盐噪声的同时能够较好地保持图像目标信息。该方法针对疑似噪声点采用二次检测的方式,并且结合改进的自适应中值滤波和能量函数模型进行灰度值恢复。该文分别使用仿真试验和真实星图处理试验对该方法进行验证,在客观评价中,图像峰值信噪比PSNR(Peak Signal to Noise Ratio)最高可提高3倍多,均方误差MSE(Mean Squared Error)减小为加噪图像的3.16×10~(-5)。试验结果表明,该方法可有效地降低传统方法的噪声误检问题,同时提高噪声图像的恢复精度,非常适合星图噪声的去除。
[Abstract]:The star map with deep space background taken by ground-based observation camera is often affected by the complex background of the star sky and often has a high noise level. At the same time, the star map is mainly composed of star, space object and star background noise, and it is a dot distribution, so the traditional image denoising algorithm is not suitable for star map. In this paper, an energy function based median filter denoising algorithm is proposed, which not only removes the salt and pepper noise of star map, but also preserves the target information of the image. The method uses quadratic detection for suspected noise points and combines improved adaptive median filter and energy function model for gray value recovery. In this paper, simulation experiments and real star map processing experiments are used to verify this method. In objective evaluation, the peak signal-to-noise ratio (PSNR (Peak Signal to Noise Ratio) of the image can be increased by more than three times. The mean square error (MSE (Mean Squared Error) is reduced to 3.16 脳 10 ~ (-5) of the noised image. The experimental results show that the proposed method can effectively reduce the problem of noise false detection and improve the accuracy of noise image restoration. It is very suitable for star map noise removal.
【作者单位】：中国科学院长春光学精密机械与物理研究所;中国科学院大学;
【基金】：国家863计划项目(2011AA8082035) 中国科学院长春光学精密机械与物理研究所三期创新工程资助项目(065X32CN60)~~
【分类号】：TP391.41

【相似文献】