Figure 1.
Illustrates reconstruction by opening of mark image relative to mask image
.
Figure 2.
Illustrates reconstruction by closing of mark image relative to mask image
.
Figure 3.
Uneven illumination text image.
Figure 4.
Result of Top-Hat transform with block effect.
Figure 5.
(a)–(b) Uneven background with block effect.
Figure 6.
(a) Original image, (b) result of conventional opening of (a) using a disk SE, and (c) result of opening by reconstruction of (a) with same SE.
Figure 7.
Background without block effect but with over illuminated.
Figure 8.
(a)–(b) Background without any block effect or over illumination.
Figure 9.
Different SEs used in proposed algorithm.
Figure 10.
Diagram of the proposed algorithm.
Figure 11.
Final illumination normalized output without block effect or fake contours.
Figure 12.
Ideal text image.
Figure 13.
(a)–(d) Four different uneven illumination modes.
Figure 14.
(a)–(d) Synthetic images with the four different uneven illumination modes.
Figure 15.
Results of histogram equalization algorithm.
Figure 16.
Results of histogram specification algorithm.
Figure 17.
Results of method proposed by[16].
Figure 18.
Results of classical TOP-HAT transform.
Figure 19.
Results of the proposed algorithm.
Figure 20.
English text image, (b) uneven illumination background extracted by Top-Hat, (c) Equalization results from Top-Hat, (d) Equalization results from proposed algorithm.
Figure 21.
Chinese text image, (b) uneven illumination background extracted by Top-Hat, (c) Equalization results from Top-Hat, (d) Equalization results from proposed algorithm.
Figure 22.
Japanese text image, (b) uneven illumination background extracted by Top-Hat, (c) Equalization results from Top-Hat, (d) Equalization results from proposed algorithm.