HAN Ying-li.A Modified Total Variation Regularization Shearlet Adaptive Algorithm for Steel Strip Image Denoising[J],43(6):105-110 |
A Modified Total Variation Regularization Shearlet Adaptive Algorithm for Steel Strip Image Denoising |
Received:June 17, 2014 Revised:December 10, 2014 |
View Full Text View/Add Comment Download reader |
DOI: |
KeyWord:cold-rolled steel strip image denoising Shearlet transform total variation model |
Author | Institution |
HAN Ying-li |
School of Mechanical Engineering, Tianjin Polytechnic University, Tianjin , China |
|
Hits: |
Download times: |
Abstract: |
Objective To effectively remove mixed noise from the image of acquisition steel strip in the production field. Methods Combining the advantage of the Shearlet transform which has better properties to sparsely express the characteristics of the images and the total variational anisotropic diffusion, a new image denoising model was proposed. After Shearle transform decomposition, the image was processed by hard thresholding, and then the estimated image was formed after Shearle transform reconstruction. The algorithm used iterative model of minimization of total variation regularization to correct the estimated image. Results The denoised image had good visual effect, and the creation of pseudo Gibbs effect was avoided. The comparison of the new model with wavelet denoising under the strong noise level showed that PSNR was increased by 9 dB and MSE was reduced by 319. Conclusion Numerical examples demonstrated that this method could achieve better PSNR gain, and the results showed that the filters had high fidelity of signal amplitude, and better function in smoothing noise and preserving edges. |
Close |