A comparative study on 3D range data compression methods

As high-quality 3D range scanners become increasingly adopted, a common issue emerges that is how best to properly store captured 3D data as it inherently contains a large amount of information per each frame. One approach that has proved successful is to convert 3D range data to 2D regular color images that can be further compressed using traditional image compression techniques (e.g., JPEG). In literature, there are three major conversion methods: (1) virtual fringe projection; (2) direct depth encoding; and (3) multiwavelength depth en- coding. This paper compares the effectiveness and limitations of all three major compression methods, especially when the resultant 2D images are stored with low-quality lossy (i.e., JPEG) image formats. Experimental data found that multiwavelength depth encoding outperforms both other methods, especially under various levels of lossy JPEG compression. Principles of each method will be explained, and experimental data will be presented to evaluate each method.