This project attempts to improve quality of images gained by digital still cameras. Resolution of current cameras is still not sufficient for some scientific purposes, but under certain circumstances it may be improved by using posprocessing methods known as Super-Resolution.


Multiple images of a single unchanged scene are taken and a single image of a higher resolution is reconstructed by improved Super-Resolution method. The improvement is based on the knowledge of the processing that takes place within current off-the-shelf digital still cameras. Furthermore, registration (i.e. process of finding the shift coefficients) was significantly accelerated using GPU hardware.



Váša, L.; Hanák, I.; Skala, V.: Improved Super-Resolution Method and Its Acceleration, Proceedings of EUSIPCO 2005
Váša, L.; Skala, V.: Resolution Improvement of Digitized Images, Proceedings of Algoritmy 2005
Váša, L.: Resolution Improvement of Digitized Images, Diploma thesis at University of West Bohemia (2004)


Example results:
Contour Lines 2D

Figure 1: One of 22 input images obtained by digital still camera

Contour Lines 2D

Figure 2: Result of Zomet SR applied on the input

Contour Lines 3D

Figure 3: Result of our improved method