Noise suppression algorithm of short-wave infrared star image for daytime star sensor

As an important development trend of star sensor technology, research on daytime star sensor technology can expand the applications of star sensor from spacecrafts to airborne vehicles. The biggest problem for daytime star sensor is the detection of dim stars from strong atmospheric background radiation. The use of short-wave infrared (SWIR) technology has been proven to be an effective approach to solve this problem. However, the SWIR star images inevitably contain stripe nonuniformity noise and defective pixels, which degrade the quality of the acquired images and affect the subsequent star spot extraction and star centroiding accuracy seriously. Because the characteristics of stripe nonuniformity and defective pixels in the SWIR star images change with time during a long term continuous operation, the method of one-time off-line calibration is not applicable. To solve this problem, an algorithm of noise suppression for SWIR star image is proposed. It firstly extracts non-background pixels by one-dimensional mean filtering. Then through one-dimensional feature point descriptor, which is used to distinguish the bright star spots pixels from defective pixels, various types of defective pixels are accurately detected. Finally, the method of moment matching is adopted to remove the stripe nonuniformity and the defective pixels are compensated effectively. The simulation experiments results indicates that the proposed algorithm can adaptively and effectively suppress the influence of stripe nonuniformity and defective pixels in SWIR star images and it is beneficial to obtain higher star centroiding accuracy.