Steps in Image Reconstruction from scan data


The object (a Vase) is rotated in front of the color laser scanner.  One profile is recorded for every degree of rotation (for a total of 360 profiles). Each profile contains 360 coordinates (x,y,z,R,G,B). Due to the simultaneous digitizing of the CCD pulse position (for the range) and amplitude (for the color), the registration of the geometric data and the color data is perfect.
"NRC first high resolution 3-D Color digitizing of a vase (1990)."

Raw data is shown in the lower left corner, the color is collected at 12 bits per color (but usually reduced to 8 bits per color) and the range data is at 2 bytes per pixel. The range file of the vase here appears unfolded due to the rotation of the object while digitizing. A coordinate  transform is applied to this file in order to reconstruct a view of the vase. In yellow is show the reconstructed data points, in red we have a wire mesh reconstruction using a subsample of the range file, in white a surface model from all the data points and, the next views show the color mapping of the raw data and a synthetic view using illumination from the right side, respectively.

Images provided by
   National Research Council
   Institute for Information Technology
   Visual Information Technology Group