Skip to main content

Table 1 Performance of the proposed denoising filter in comparison with other ADFs

From: Geometric nonlinear diffusion filter and its application to X-ray imaging

Images

Matrix size

NERR

NMSE

SNR gain

β

Computation time (s)

Shepp-Logan

Phantom

512 × 512

0.979

(0.731)

[0.927]

0.0885

(0.364)

[0.196]

3.40

(2.43)

[2.34]

0.987

(0.858)

[0.964]

0.428

(1.050)

[0.863]

Mouse

Chest

2240 × 2240

0.942

(0.904)

[0.903]

0.206

(0.393)

[0.488]

2.10

(1.80)

[1.74]

0.148

(0.271)

[0.113]

8.072

(19.17)

[15.45]

Micro-CT

Sagittal

512 × 512

  

3.46

(2.92)

[3.26]

0.809

(0.886)

[0.986]

0.484

(1.26)

[0.781]

Micro-CT

Coronal

1024 × 1024

  

2.07

(2.05)

[2.00]

0.894

(0.914)

[0.902]

1.723

(3.989)

[3.139]

Micro-CT

Axial

512 × 512

  

2.19

(1.75)

[1.36]

0.736

(0.803)

[0.936]

0.470

(1.01)

[0.949]

Human

Chest

3072 × 3072

0.782

(0.778)

[0.747]

0.376

(0.437)

[0.470]

1.79

(1.65)

[1.59]

0.163

(0.174)

[0.166]

12.94

(34.69)

[26.57]

  1. ( ) represents the results obtained with Weickert ADF.
  2. [ ] represents the results obtained with Perona-Malik ADF.