Table 3 SSIM comparison with published results for several competitive denoising algorithms using five different sequences and three noise standard deviation levels

10

SNLM

0.887

0.853

0.934

0.890

0.907

0.894

BM3D

0.917

0.869

0.963

0.959

0.918

0.925

WRSTF [31]

0.932

0.897

0.953

0.908

0.927

0.923

SEQWT [32]

0.900

0.842

0.941

NA

NA

0.894

3DWTF [33]

0.923

0.856

0.909

NA

NA

0.896

IFSM [34]

0.904

0.855

0.927

0.904

0.886

0.895

3DSWDCT [36]

0.955

0.894

0.959

0.946

0.932

0.937

VBM3D

0.959

0.916

0.963

0.967

0.934

0.948

ST-GSM [36]

0.960

0.894

0.950

0.952

0.937

0.939

DNLM [16]

0.931

0.856

0.947

0.964

0.946

0.928

RNLM (no BMA)

0.937

0.881

0.953

0.954

0.923

0.930

RNLM (BMA)

0.939

0.895

0.960

0.960

0.925

0.936

15

SNLM

0.825

0.759

0.886

0.829

0.870

0.834

BM3D

0.878

0.817

0.937

0.948

0.888

0.893

WRSTF [31]

0.901

0.839

0.922

0.877

0.877

0.883

SEQWT [32]

0.846

0.722

0.893

NA

NA

0.820

3DWTF [33]

0.903

0.793

0.872

NA

NA

0.856

IFSM [34]

0.851

0.776

0.882

0.857

0.836

0.840

3DSWDCT [36]

0.930

0.834

0.931

0.928

0.907

0.906

VBM3D

0.943

0.874

0.940

0.961

0.911

0.926

ST-GSM [36]

0.941

0.841

0.925

0.943

0.917

0.913

DNLM [16]

0.889

0.795

0.906

0.951

0.929

0.894

RNLM (no BMA)

0.883

0.811

0.900

0.941

0.887

0.884

RNLM (BMA)

0.902

0.829

0.933

0.945

0.901

0.902

20

SNLM

0.768

0.679

0.836

0.761

0.833

0.775

BM3D

0.843

0.780

0.909

0.936

0.861

0.866

WRSTF [31]

0.868

0.790

0.889

0.846

0.873

0.853

SEQWT [32]

0.796

0.716

0.842

NA

NA

0.785

3DWTF [33]

0.882

0.740

0.840

NA

NA

0.821

IFSM [34]

0.801

0.709

0.837

0.812

0.793

0.790

3DSWDCT [36]

0.905

0.790

0.900

0.909

0.884

0.878

VBM3D

0.923

0.836

0.918

0.956

0.891

0.905

ST-GSM [36]

0.923

0.797

0.900

0.936

0.901

0.891

DNLM [16]

0.849

0.758

0.865

0.939

0.913

0.865

RNLM (no BMA)

0.881

0.779

0.881

0.930

0.855

0.865

RNLM (BMA)

0.886

0.783

0.905

0.937

0.876

0.877