Curvelet‐based multiscale denoising using non‐local means &amp; guided image filter

Panigrahi, Susant Kumar; Gupta, Shipra; Sahu, Prasanna Kumar

doi:10.1049/iet-ipr.2017.0825

Cited by 30 publications

(34 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [25], author incorporated the advantages of both (multiscale) NLM filtering and hard thresholding in three different scales of the curvelet: the approximation, the coarser and the fine scale. The edge preserving the property of the non-linear NLM filter ensures the suppression of noise in the approximation scale and aids in preserving well connected edges with small image details in the fine scale.…”

Section: = +mentioning

confidence: 99%

“…The results of this on both grayscale and colour images have shown encouraging quantitative and qualitative improvement compared with several state-of-the-art techniques at higher noise strength. However, this technique unable to preserve the finescale details at lower noise strength and it is quite hard to sampling on a grid of rectangular for the transform presented in [25] since it was implemented in continuous and the directions apart from horizontal and vertical are very unlike on the grid of rectangular.…”

Section: = +mentioning

confidence: 99%

“…The denoising quality of the proposed algorithm was evaluated and compared using PSNR (in dB) and SSIM indices. CT [12], NLM filtering [3], MBF [13], NLM method noise thresholding (NLMNT) [17] and the state-of-the-art BM3D [19], GSRC [21] and CT-NLM-GF [25] are considered in denoising quality comparison for the grayscale image. Fig.…”

Section: Visual Analysismentioning

confidence: 99%

“…7 respectively. Figure. 4 Denoised output images of 'Lena' using the: (a) CT [12], (b) NLM [3], (c) MBF [13], (d) NLMNT [17], (e) BM3D [19], (f) GSRC [21], (g) CT-NLM-GF [25], and (h) proposed denoising model with = 10 Figure. 5 Denoised output images of 'Lena' using the: (a) CT [12], (b) NLM [3], (c) MBF [13], (d) NLMNT [17], (e) BM3D [19], (f) GSRC [21], (g) CT-NLM-GF [25], and (h) proposed denoising model with = 20…”

Section: Visual Analysismentioning

confidence: 99%

“…Denoised output images of 'Lena' using the: (a) CT [12], (b) NLM [3], (c) MBF [13], (d) NLMNT [17], (e) BM3D [19], (f) GSRC [21], (g) CT-NLM-GF [25], and (h) proposed denoising model with = 30 Figure. 7 Denoised output images of 'Lena' using the: (a) CT [12], (b) NLM [3], (c) MBF [13], (d) NLMNT [17], (e) BM3D [19], (f) GSRC [21], (g) CT-NLM-GF [25], and (h) proposed denoising model with = 40 Table 1 demonstrates that the quality evaluation parameters like PSNR and SSIM for the denoising algorithms like CT [12], NLM [3], MBF [13], NLMNT [17], BM3D [19], GSRC [21], CT-NLM-GF [25] and proposed denoising model with the values ranging from 10 to 40, where our proposed model obtained the best values of both PSNR and SSIM which are highlighted in bold letter for visual convenience. However, here mean PSNR is considered instead of individual image PSNR which is further compared with conventional denoising approaches as demonstrated in Table 1.…”

Section: Visual Analysismentioning

confidence: 99%

See 4 more Smart Citations

An Effective Method for Image Denoising Using Non-local Means and Statistics based Guided Filter in Nonsubsampled Contourlet Domain

Bhargava¹,

Sivakumar²

2019

IJIES

View full text Add to dashboard Cite

This article presents an effective method for image denoising using non-local means with packing of multi patches (NLM-PMP) in non-subsampled contourlet (NSC) domain which works based on multi-scale decomposition and directionality. Guided filter with image statistics (SGIF) is utilized to further process the obtained denoised image for mitigation of ringing artifacts those are invariably appeared, which leads to the local structure preservation like textures, edges and small details efficiently. The authors decomposed the image into low-pass subband coefficients and bandpass subband coefficients including the base and the detail scales. The low frequency noise in the base sub-band and the edges with small textural details in the detail scale are processed independently using an NLM-PMP filter. In addition, SGIF is applied to enhance the visual perception of denoised image. Experimental results indicate that the proposed approach is competitive at lower noise strength with respect to peak signal to noise ratio and structural similarity index measure and excels in performance at higher noise strength compared with several state-of-the-art algorithms.

show abstract

Section: = +mentioning

confidence: 99%

Section: = +mentioning

confidence: 99%

Section: Visual Analysismentioning

confidence: 99%

Section: Visual Analysismentioning

confidence: 99%

Section: Visual Analysismentioning

confidence: 99%

See 3 more Smart Citations

An Effective Method for Image Denoising Using Non-local Means and Statistics based Guided Filter in Nonsubsampled Contourlet Domain

Bhargava¹,

Sivakumar²

2019

IJIES

View full text Add to dashboard Cite

show abstract

Curvelet and fast marching method‐based technique for efficient artifact detection and removal in dermoscopic images

Choudhary

Singhai

Yadav

2021

Int J Imaging Syst Tech

View full text Add to dashboard Cite

Effective detection and removal of the artifacts from dermoscopic images require the analysis of structures within lesions for accurate cancer detection. But the dermoscopic images acquired may be affected by image acquisition noise, variation in texture, color, and irrelevant structure such as hairs, air bubbles, and contour blurring around the lesion. Hence preprocessing is very essential for accurate diagnosis. The proposed algorithm includes the preprocessing, to remove acquisition noise using curvelet transform, illumination correction using contrast limited adaptive histogram equalization (CLAHE) algorithm for image enhancement, Frangi vesselness algorithm is utilized for detection and removal of hairs and vessels like structures present in enhanced image, fast marching inpainting method is applied for repairing the information of lesion from removed hair pixels. The performance of the proposed algorithm is analyzed based on commonly used parameters such as diagnostic accuracy (DA), sensitivity (SE), specificity (SP), and precision (P). The performance is evaluated on PH2 and International Symposium on Biomedical Imaging (ISBI) datasets than compared with existing methods. The proposed algorithm segments the skin lesion with the highest diagnostic accuracy of 81.49% and sensitivity of 93.88% for the PH2 dataset and 75.39% and 79.34% respectively for the ISBI dataset. Results are better than existing methods. The proposed algorithm with improved preprocessing for artifact detection and removal is highly accurate and able to retrieve the hair occluded information.

show abstract