Skip to main content

Table 5 Hardware complexity in DSP context

From: A robust anisotropic diffusion filter with low arithmetic complexity for images

Methods

Noise type

Noise %

λ

NIAP

No. of A/M/S per iteration (C)

Running time per iteration in seconds

Running time for NIAP in seconds*

Computational efficiency of FORADF (Cref) in comparison with RF and AMD in terms of excess/savings on C (%)

((C − Cref)/C)

RF

GN + S&P

0.1 + 70

0.25

100

11/18/0

3.5244

352.44

((CRF − Cref)/CRF)= 41.37% excess

AMD

GN + S&P

0.1 + 70

0.25

100

12/33/24

5.5495

554.95

((CAMD − Cref)/CAMD)= 75.36% excess

FORADF

GN + S&P

0.1 + 70

1

5

5/4/8

4.9407

24.7038

((Cref − Cref)/Cref)= 0% excess

  1. A/M/S additions/multiplications/sorting, NIAP number of iterations required for acceptable performance, C arithmetic complexity
  2. *MATLAB R2014a, 567u7 Intel (R) core (TM) i5-5200 U CPU @ 2.20 GHz, 64—bit Operating System, RAM-4.00 GB