Our image sequence is taken by rotating and translating the camera.
But the images are in order. Therefore, we calculated transformation
parameters between each successive images. When we calculated all
transformation parameters, then we used them in blending operation.
For each image pair, we could get very good results but when we use
these parameter in blending, it could cause some artifacts.
We calculate the transformation parameters between any successive image
then we use them to construct mosaicing.
we got very good result when we blended two,three or four images into
one. But if we increase the number of images, the number of artifacts
also increases.
Therefore, we proposed different blending approach, tree-based
mosaicing.
As the transformation error between two images is less, instead of
calculating transformation parameters once and use them construct a
mosaic, we calculate the transformation parameter between image
one-two, three-four, etc. Then we use transformation parameters for
image 1 and image 2 to construct a submosaic.
For each image pair we get a submosaic.
After construct the 
submosaic where n is the number of images,
we use same method, we take two successive submosaic and we calculate
the transformation parameters between these submosaic pairs.
We continue this process until we get only one mosaic.
In figure 2, illustration of tree-based and normal mosaicing is given.
Figure 2: Tree-based image Mosaicing