一种改进的星图降噪算法研究

在星敏感器的使用过程中，由于外界环境的影响及传感器自身的限制，拍摄出来的星图不可避免地存在一些噪声，因此对星图进行去噪处理是一项非常重要的工作。针对传统高斯模板滤波存在的引入邻域噪声、无法自行根据星图特性修正等造成去噪效果不好的问题，提出了一种改进的星图降噪算法。该方法在滤波前先进行坏点剔除工作，并采用高斯低通滤波与高通滤波结合的方式对图像进行处理，在抑制噪声的同时有效地保留了星点信号。通过阐述星敏感器的工作原理，分析星图的噪声特性，对星图滤波去噪算法进行研究，并进行模拟星图影像提取星点坐标实验。结果表明：使用该算法进行滤波比传统的高斯滤波算法提取的质心坐标精确度更高，较传统方法横坐标提高0.00538个像素点，纵坐标提高0.0077个像素点，证实了图像处理算法的有效性。

Research on an Improved Denoising Algorithm for Star Image

During the use of star sensors, due to the influence of external neutrality and the sensor''s own restrictions, some noise will inevitably exist, so that denoising treatments for the star image are very important work. Aiming at the problems of the traditional Gaussian template filtering that the noise removal effect is not good due to the introduction of neighborhood noise and the inability to correct the characteristics of the star map, a new noise reduction algorithm is proposed. This method first removes the dead pixels before filtering, and uses the combination of Gaussian low-pass filtering and high-pass filtering to process the image, which effectively retains the star signal while suppressing noise. This paper explains the working principle of the star sensor, and analyzes the noise characteristics of star images; the star map filtering and denoising algorithms are studied, and tests of extracting the star point coordinates from the simulated star images are conducted. The results show that the centroid coordinates extracted by this algorithm are more accurate than the traditional Gaussian filtering algorithm. The accuracy of coordinate is increased by 0.00538 pixels, and the accuracy of ordinate is increased by 0.0077 pixels, which verifies the effectiveness of the image processing algorithm.