MI Based Registration of Multi-Sensor and Multi-Temporal Images
191
a
b
Fig. 7.7a,b. Two small portions of Landsat TM band 1 images taken in a Year 1995 and b Year
1998, respectively. The white box in image b indicates the corresponding sub-set of image a
(128 x 128 pixels and 228 x 228 pixels) of Landsat TM band 1 images taken in
years 1995 and 1998 (see Fig. 7.7). The images are systematically corrected, and
thus contain no radiometric errors. From this figure, a significant change between the intensity values of the two images can be observed which is probably
due to the change in crop types in the region. The 2D registration functions
obtained by using MSD and Nee as similarity measures, to register images in
Fig. 7.7, are shown in Fig. 7.8a,b. Ifboth MSD and Nee were suitable similarity
measures in this case, we would expect a global minimum in Fig. 7.8a and
a
i 0.2
~ 0.1
§ 0
~
~ -0.1
S -0.2
20
20
20
b
Fig.7.8a,b. Two dimensional registration functions while registering images shown in
Fig. 7.7 using a MSD and b NCC as similarity measures
191
a
b
Fig. 7.7a,b. Two small portions of Landsat TM band 1 images taken in a Year 1995 and b Year
1998, respectively. The white box in image b indicates the corresponding sub-set of image a
(128 x 128 pixels and 228 x 228 pixels) of Landsat TM band 1 images taken in
years 1995 and 1998 (see Fig. 7.7). The images are systematically corrected, and
thus contain no radiometric errors. From this figure, a significant change between the intensity values of the two images can be observed which is probably
due to the change in crop types in the region. The 2D registration functions
obtained by using MSD and Nee as similarity measures, to register images in
Fig. 7.7, are shown in Fig. 7.8a,b. Ifboth MSD and Nee were suitable similarity
measures in this case, we would expect a global minimum in Fig. 7.8a and
a
i 0.2
~ 0.1
§ 0
~
~ -0.1
S -0.2
20
20
20
b
Fig.7.8a,b. Two dimensional registration functions while registering images shown in
Fig. 7.7 using a MSD and b NCC as similarity measures
