Programming Assignment 2: View Transformation

  Liu Ren
 CSD Ph.D. student
Wean Hall 8205

Original image:


Holes and Hole Filling:

The following images are generated by view transformation.I use McMillan's marks to identify the following images as
A,B,C,D,E,F,G,H, I.So the first image(Image A)  corresponds to Region A and w>0 's case in McMillan's
paper,and so on. All the images are not resized.

Image A and Image B:



Image C and Image D:

Image E(w>0) and Image E(w<0):


Image F and Image G:

Image H and Image I(w<0):


Image  F(w<0):

A Profile View of the Eagle:


About My Work:

1.I use triangle surface model to perform the warping process.Triangles are chosen according to warp direction in each 2*2
block.Using two kinds of color interpolation methods to fill the holes.

2.I provide two ways for users to locate the camera. One way, the user can change the camera's location and rotation freely. The other
way, the camera always points to the center of the object.It can move around the object and can zoom in or zoom out. Thus the objects
are always appeared in the user's "eyes".

3.I use VC6.0++ and OpenGL to do all the coding. However, I only use the OpenGL's raster graphics feature ( glDrawPixels) to draw
points.It takes 3-5 seconds to generate a new picture.

  Liu Ren    10/3/99