This morning I spend a few hours (while I am watching TV program)to implement the algorithm in the paper “Color Map Optimization for 3D Reconstruction with Consumer Depth Cameras” .

It is very simple to implememt. The author din’t provide the raw data for test, I will test it a few days later when we captured some high quality test data. The result of the paper looks like this:

This is a test for my implementation.

From the botton picure, we can see the color map has been optimized resulting better fitting of feature points. Although the result is not good as the paper. I think it may be due to the poor quality captured images (the resolution of image is low (640480) ,the paper used high resolution (12801024) images. Some of them we captured are blured,we should removed these images) and the eye sometimes open and sometimes closed (always moving not a static object).

You can download my program from here. But you must email me so that I can tell you how to run this program. If someone need this in your code,you can ask me to send you an API library.