全变差正则化的Shearlet收缩去噪.pdfVIP

16 2 Vol. 16, No. 2 2011 2 Journal of Image and G raphics Feb. , 2011 : TN 11. 73 : A : 1006-8 61( 2011) 02-0168-06 : , , , , . Shearlet [ J]. , 2011, 16( 2) : 168- 173 Shearlet 1) 1), 2) 2) 2) 1) 胡海智 , 孙辉 , 邓承志 , 陈习 , 柳枝华 1) 2) (, 330063) (, 3300 ) : Shearlet, Shearlet, Shearlet, Shearlet Shearlet , , , PSNR : Shearlet; ; ; Sh earlet shrinkage de-noising based total variation regu larization 1) 1) , 2) 2) 2) 1) Hu H aizhi , Sun Hu i , Deng Chengzh i , Chen X i , Liu Zh ihua 1) (School of Inf ormation Science and E ng ineering, N anchangH angkong University, N anchang 330063 China) 2) (D epartment of Comp uter Science and echnology, N anchang Institute of echnology, N anchang 3300 China) A bstrac t: Shearlet is a new-style m ulti-scale geom etry analysis tool. It creates Shearlet functions wh ich have d ifferent characteristics through zoom ing, shearing, translating and other affine transform ing m ethods, and enables its capab le of optmi ally sparse representation. Firstly, a d igital Shearlet transform mi plem entation m ethod is proposed in th is paper. And then, a new de-noising m ethod that comb ines Shearlet transform and variation is presented, wh ich m ainly been established using a total variation regu larization m odel to constraint condition on Shearlet transform domain. The proposed m odel ami s at reducing Pseudo-G ibbs artifacts after smi p le threshold m ethods. Num erical exam ples dem onstrate that the m e