图像降噪与星点识别是弹载星敏感器进行星点质心提取的前提,针对其昼间星点识别困难的问题,提出了一种基于小波变换的去噪声方案。该方案选择双正交小波函数对图像进行分解和软阈值化处理。在此基础上,进行图像信号重建,得到降噪后的图像。对复原后的图像采用改进的最大类间方差法进行了星点识别。星点识别的准确率达到了86.7%。实验结果证明了此方案在北极星昼间测量与识别上的可行性。

Image de-noising and star-recognizing are prerequisites to centroid algorithm. In view of the centroid problem, the solution adopts bi-orthogonal 3.7 as the mother wavelet to conduct decomposition in multiple levels. A Soft-threshold algorithm was employed on each high frequency level of image signal to suppress the noise components and guarantee the smoothness. Subsequently, the signal components of all levels were used to reconstruct the de-noised image. Then, a modified OTSU algorithm was performed for the star recognition. According to the experiment, the accuracy of the star recognition accounts for 86.7%. The result demonstrates that the solution is feasible in daytime Polaris observation and recognition.