SPARSE RECONSTRUCTION METHOD BASED ON STARLET TRANSFORM FOR HIGH NOISE ASTRONOMICAL IMAGE DENOISING

摘要

High noise astronomical image denoising has been the bottleneck and key point in deep space exploration. Designing an effective astronomical image denoising method plays a crucial role in analyzing astronomical image details. The famous compressed sensing (CS) has been proved to be a successful technique for high dimensional signal. In this paper, CS is adopted to address the denoising problem of high noise astronomical images, and a CS recovery model and an iterative starlet shrinkage thresholding (ISST) reconstruction algorithm are put forward simultaneously. In l_1 norm CS recovery model, the internal structure of astronomical images is fully considered, which affects the result of image reconstruction. Combining l_1 norm and fractional-order total variation (FRTV), an improved CS reconstruction model is first established to preserve more astronomical image details. In this framework, an adaptive starlet thresholding operator proposed in ISST algorithm is used to select these sparse coefficients in the process of astronomical image starlet sparsity transform. Moreover, an optimized BayesShrink thresholding is developed to pick out the reconstructed astronomical image data in each iteration. The results of various experiments consistently show that the algorithm proposed can efficiently recover high quality astronomical images, and preserve more astronomical image details. Therefore, this algorithm can be applied in the ground receiving station to accurately recondstructing high quality images from high noise astronomical images.