Research Article

A “Salt and Pepper” Noise Reduction Scheme for Digital Images Based on Support Vector Machines Classification and Regression

Figure 10

Reconstruction results for the Lenna and Barbara images with 50% and 90% of “salt and pepper” noise, using the MDBUTMF (50%), MMEM (90%), SVM-M2, and SVM-R methods.
826405.fig.0010a
(a) Lenna 50%
826405.fig.0010b
(b) MDBUTMF (30.95 dB)
826405.fig.0010c
(c) SVM-M2 (30.90 dB)
826405.fig.0010d
(d) SVM-R (31.06 dB)
826405.fig.0010e
(e) Barbara 50%
826405.fig.0010f
(f) MDBUTMF (25.45 dB)
826405.fig.0010g
(g) SVM-M2 (25.20 dB)
826405.fig.0010h
(h) SVM-R (26.40 dB)
826405.fig.0010i
(i) Lenna 90%
826405.fig.0010j
(j) MMEM (23.96 dB)
826405.fig.0010k
(k) SVM-M2 (24.13 dB)
826405.fig.0010l
(l) SVM-R (20.33 dB)
826405.fig.0010m
(m) Barbara 90%
826405.fig.0010n
(n) MMEM (20.92 dB)
826405.fig.0010o
(o) SVM-M2 (20.54 dB)
826405.fig.0010p
(p) SVM-R (19.31 dB)