Research Article

Image Deconvolution by Means of Frequency Blur Invariant Concept

Table 3

Image deconvolution using proposed frequency blur invariant for Tropical Storm Lorenzo image (shown in Figure 3(a)) with different estimated (below the restored images), different mask sizes, and their corresponding NMSE.

Blurred
image (A)
Deblurred images

Mask size:
951842.table.0011951842.table.0012951842.table.0013951842.table.0014951842.table.0015951842.table.0016951842.table.0017951842.table.0018951842.table.0019
0.25 0.73 1.34 1.98 2.65 3.12 3.28 3.45

NMSE 0.4429 0.4106 0.3619 0.3007 0.2491 0.1838 0.0195 0.0056

Mask size:
951842.table.0020951842.table.0021951842.table.0022951842.table.0023951842.table.0024951842.table.0025951842.table.0026951842.table.0027951842.table.0028
0.42 0.87 1.68 2.56 4.98 7.24 7.51 7.72

NMSE 0.7915 0.6745 0.5608 0.4834 0.1973 0.0088 0.0031 0.0015

Mask size:
951842.table.0029951842.table.0030951842.table.0031951842.table.0032951842.table.0033951842.table.0034951842.table.0035951842.table.0036951842.table.0037
1.53 2.87 3.19 4.01 6.79 8.26 10.93 11.28

NMSE 0.8924 0.7127 0.5111 0.3501 0.1833 0.1381 0.0551 0.0110