Abstract
Truncated singular value decomposition (TSVD) is a simple and efficient technique for patch-based image denoising, in which a hard thresholding operator is utilized to set some small singular values to zero. Before performing the hard thresholding, the noise variance should be accurately estimated in order to determine the rank of the patch matrices. However, when the noise level is high, the denoisied results from the TSVD still contain some residual noise. To solve this problem, we present a hybrid thresheldoing strategy that combines a hard thresholding operator and a soft one. The former is directly reused the thresholding derived from TSVD, the latter is derived by minimum variance estimator. Simulation experiments are conducted to verify the effectiveness of the proposed method. Experimental results show that the method can effectively denoise the images with high level noise.
Highlights
During the acquisition, storage and transmission of images, there is inevitably noise due to factors such as equipment, environment and long sampling time [1]
PROPOSED METHOD In order to improve the performance of low rank approximation singular value decompostion (SVD) (LRA-SVD), we apply the minimum variance estimate theory to further estimate each singular value preserved by the hard thresholding operator, and yield a soft shrinkage operator
The basic parameters used in these three methods include: image patch size, the number of similar patches in each patch matrix, the projection constant used in iterative back projection, the scaling factor used in noise level updating, and the number of iterative back projection
Summary
Storage and transmission of images, there is inevitably noise due to factors such as equipment, environment and long sampling time [1]. If the self-similar pattern can not be found from the noisy image itself, one can use other similar but noise-free images This idea leads to another class of denoising methods [20]– [22] that use Gaussian mixture model to learn natural images to create a prior condition, which is used to constrain the results that need to be estimated. Deep learning based denoising algorithms, e.g., DnCNN [23] and FFDNet [24], learn a multilayer network structure for image representation through a large amount of data, in which external prior information can be learned. In [25], Guo et al proposed a new denoising model called CBDnet, which consists of two parts: the noise estimation sub-network and the denoising sub-network It performs a joint training based on signal-independent noise and noisecombined images. This improves the generalization ability of the denoising network and enhances its denoising performance
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
