[english]
[français]

Probabilistic Patch-Based filter (PPB)

[Previous] [Next] [Homepage]

Description of the filter

Algorithme NL means
The NL means algorithm computes the mean for all pixel s of the values of the pixels t weighted with respect to the similarity between two windows centered respectively on s and t
equation
equation
up

Qualitative Evaluation of the Denoising Algorithms

lena: awgn_noisy(a)
lena: awgn_bm3d(b)
lena: awgn_ppbit25(c)
(a) From top to bottom, corrupted images of Lena, by an additive WGN with standard deviation σ = 40. Denoised images using (a) Noisy (b) BM3D (c) PPB 25 it.
lena: mgsn_noisy(a)
lena: mgsn_newwavmap(b)
lena: mgsn_ppbit25(c)
(a) From top to bottom, corrupted images of Lena, by a multiplicative GSN with equivalent number of looks L = 3. Denoised images using (a) Noisy (b) MAP-UWD-S (c) PPB 25 it.
bayard_L1: mgsn_noisy(a)
bayard_L1: mgsn_newwavmap(b)
bayard_L1: mgsn_ppbit25(c)
(a) From top to bottom, SAR images of Bayard ©DGA ©ONERA, by a multiplicative GSN with equivalent number of looks L = 3. Denoised images using (a) Noisy (b) MAP-UWD-S (c) PPB 25 it.

Show more comparisons and results

Synthetic images:
Noise-free image:
Additive WGN:
Multiplicative GSN:
Real SAR images:
Noisy image:
Denoising methods:
Additive WGN:
Multiplicative GSN:
up

Quantitative Evaluation of the Denoising Algorithms

TABLE I
SNR values of estimated images using different denoising methods for images corrupted by (left) an additive WGN with different standard deviations and by (right) a multiplicative Goodman's speckle noise with different equivalent number of looks
σ = 10 σ = 20 σ = 40 σ = 60
Barbara
Noisy image 14.738.8043.0900.043
   
   
BLS-GSM 19.7315.7611.549.083
K-SVD 21.0217.4313.019.285
BM3D 21.4818.3814.5912.14
NL means19.8516.9712.8510.24
PPB 25 it 18.6915.9613.4910.99
Boat
Noisy image 13.417.4241.627-1.490
   
   
BLS-GSM 18.7315.6012.099.548
K-SVD 18.8715.6211.789.039
BM3D 19.0916.0912.8310.55
NL means17.5914.6311.068.959
PPB 25 it 17.1914.5111.63 9.503
House
Noisy image 13.277.2631.445-1.617
   
   
BLS-GSM 20.5317.7314.2811.52
K-SVD 21.1518.3114.3610.22
BM3D 21.7718.9415.7813.28
NL means20.2517.5513.3310.40
PPB 25 it 19.5917.0314.2011.57
Lena
Noisy image 13.597.5971.814-1.251
   
   
BLS-GSM 20.6717.6614.4811.69
K-SVD 20.9317.8114.1811.09
BM3D 21.2718.4215.3313.05
NL means20.1217.1013.6611.33
PPB 25 it 19.5016.9014.2011.99
L = 1 L = 2 L = 4 L = 16
Barbara
Noisy image-1.0901.6944.61110.57
Kuan 6.9378.48910.1914.41
WIN-SAR 8.81910.4812.0415.82
LLMMSE-UWD 9.07210.8312.6716.75
MAP-UWD 9.30110.7312.7517.00
MAP-UWD-S 9.64511.4413.2816.93
PPB non-it. 9.78511.8814.0517.83
PPB 25 it 10.5812.5113.9816.59
Boat
Noisy image -2.992-0.1792.6968.667
Kuan 6.1717.9509.79613.77
WIN-SAR 8.56810.6512.1415.17
LLMMSE-UWD 8.20410.0311.7115.45
MAP-UWD 9.27210.7812.1815.74
MAP-UWD-S 9.25710.6812.3115.71
PPB non-it. 8.70810.4912.2215.33
PPB 25 it 9.42610.9112.2515.10
House
Noisy image -3.549-0.7602.1138.096
Kuan 5.9467.8409.71213.89
WIN-SAR 8.68911.4213.1516.24
LLMMSE-UWD 8.19110.2612.1516.35
MAP-UWD 10.2211.9013.4917.07
MAP-UWD-S 10.3411.9713.7217.24
PPB non-it. 9.06411.6114.2918.27
PPB 25 it. 10.4612.9814.5017.42
Lena
Noisy image -2.4490.3433.2499.188
Kuan 7.3509.20111.2215.54
WIN-SAR 10.3513.0014.7217.90
LLMMSE-UWD 9.77511.8113.7517.80
MAP-UWD 11.7113.3014.9918.47
MAP-UWD-S 11.8713.5315.1418.65
PPB non-it. 11.0513.2015.1818.61
PPB 25 it. 12.1613.9515.2518.10
 
up

PPB software

Download the PPB filter

We recommend to use the more recent NL-SAR technique for speckle noise reduction (available here: NL-SAR)

 

These pieces of Matlab software are based on C++ Mex-Functions compiled for Linux 32-bit, Linux 64-bit and Windows 32 bit. Matlab script exemples are given, they have been written for MATLAB with the Image Processing Toolbox (to load the images). Please refer to the REAME file for more details. For any comment, suggestion or question please contact Charles-Alban Deledalle at deledalle (at) telecom-paristech (dot) fr.

up

References

  1. Goodman, JW.
    Some fundamental properties of speckle,
    J. Opt. Soc. Am, vol. 66, no. 11, pp. 1145-1150, 1976.
  2. Buades, A. and Coll, B. and Morel, J.M.
    A Non-Local Algorithm for Image Denoising,
    IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 2, 2005
  3. Portilla, J. and Strela, V. and Wainwright, MJ and Simoncelli, EP
    Image denoising using scale mixtures of Gaussians in the wavelet domain,
    IEEE Transactions on Image Processing, 2003
  4. Aharon, M. and Elad, M. and Bruckstein, A.
    K-SVD: An algorithm for designing overcomplete dictionaries for sparse representation,
    IEEE Transactions on Signal Processing, 2006
  5. Dabov, K. and Foi, A. and Katkovnik, V. and Egiazarian, K.
    Image denoising by sparse 3-D transform-domain collaborative filtering,
    IEEE Transactions on Image Processing, 2007
  6. Kuan, DT and Sawchuk, AA and Strand, TC and Chavel, P.
    Adaptive noise smoothing filter for images with signal-dependent noise,
    IEEE Transactions on Pattern Analysis and Machine Intelligence, 1985
  7. Achim, A. and Tsakalides, P. and Bezerianos, A.
    SAR image denoising via Bayesian wavelet shrinkage based on heavy-tailed modeling
    IEEE Transactions on Geoscience and Remote Sensing, 2003
  8. Argenti, F. and Alparone, L.
    Speckle removal from SAR images in the undecimated wavelet domain,
    IEEE Transactions on Geoscience and Remote Sensing, 2002
  9. Argenti, F. and Bianchi, T. and Alparone, L.
    Multiresolution MAP despeckling of SAR images based on locally adaptive generalized Gaussian pdf modeling
    IEEE Transactions on Image Processing, 2006
  10. Bianchi, T. and Argenti, F. and Alparone, L.
    Segmentation-Based MAP Despeckling of SAR Images in the Undecimated Wavelet Domain
    IEEE Transactions on Geoscience and Remote Sensing, 2008
up

Last modified: Fri Apr 7 01:23:58 Europe/Berlin 2017