Toy

Multichannel color input images originate from a digital camera.

Input pictures show different regions of the same image which are in focus. There is a background in focus on the first image and a toy on the second image.

first input image	second input image

The input images depict a rather complicated scene with a lot of different smooth and glossy objects in the background. This observation forced us to choose the CA and not the simpler SA. The resulting (fused) toy seems to be perfect except for the one blurred area that is still blurred in all the input images. This blurred area is situated in a background area that contains both smooth and glossy elements and is thus very sensitive to any disturbances. Obviously, the CA with a higher value of n_start,m_start gives a better fused image.

CA with nstart,mstart=1,kmax=10,step=2	CA with nstart,mstart=10,kmax=4,step=2

© 2007 Institute for Research and Applications of Fuzzy Modeling, University of Ostrava, Czech Republic