"Two-channel depth encoding for 3D range geometry compression," Appl. Opt. (2019)

M.G. Finley and T. Bell, "Two-channel depth encoding for 3D range geometry compression," Appl. Opt. 58(25), 6882-6890 (2019). (Cover Feature)

Abstract

This paper presents a novel method for accurately encoding 3D range geometry within only two channels of a 2D RGB image using a two-frequency phase-shifting approach. Once encoded within a 2D image, 3D geometry can be further compressed with conventional lossless or lossy image compression methods. The nature of the proposed two-channel encoding is relatively smooth; thus, large compression ratios with high reconstruction accuracies can be achieved and are experimentally demonstrated. For example, a compression ratio of 2883:1 was achieved, compared with the STL format, with a reconstruction RMS error of 0.45 mm (99.8% accuracy) when JPEG 85 was used with the proposed method. This paper also demonstrates how a 24-bit color texture map can be encoded alongside 3D geometry within a single 2D image.