Space image de-noising through getting the district where the targets possibly exist

The image taken by the visible sensor carried by star is featured by little difference between target and noise,no atmospheric attenuation,and influenced by spherical radiation greatly.The de-noising to space images requires that the target edges be kept integrally while keeping its inner gray scale distribution unchanged.Traditional de-noising methods may greatly affect subsequent work of target recognition.Thus a method is proposed which reduces the noise through getting the district where the targets possibly exist in.The method makes use of SUSAN filtering,frequency domain analysis,and background clutter reduction to get the target district,then extracts the image that has no background noise and keeps original target brightness distribution from the original image,and thus the image de-noising is realized.To verify its de-noising effect,it is compared with median filtering,low-pass filtering,and wavelet transform from three aspects.The result shows that the method can keep the target edges and inner gray scale distribution well while reducing the background noise.