[ back]

Points corresponding in integer accuracy, regular sampling with noise (noise amplitude = 1 pixel).
the noise function we use:

2*noise amplitude
*rand - 0.5.

(the rand function gives random numbers chosen from

a uniform distribution on the interval (0.0,1.0)).

Input:

Camera position view:

VRML code here

camera parameters and corresponding points
here.

Source image 1 (red camera)

Source image 2 (green camera)

Desired image (blue camera)

Output:

Corresponding points' position estimation (calculation error is 0.0225)

P3 - original points of the third image | P3B - calculated points | both P3 and P3B are on one plot; if you don't see red points, it means that the placement is very good |

Delaunay triangulation of the points generated by the 8-points algorithm.

Delaunay triangulation of the desired image applied to source images 1 and 2, respectively.

The desired Image received from source images 1 and 2, respectively.

VRML code here | VRML code here |

The first decision map on how to merge
the results we get from source images 1 and 2 :

The second decision map showing how to merge the results we get from source images 1 and 2 :

red - take current triangle's texture from source 1.
green - take current triangle's texture from source 2.
white - can't take texture from any sources. |

The desired image received both by source images 1 and 2 both .

VRML code here

[ back]