Two-channel depth encoding for 3D range geometry compression.

Area of Science:

• Computer Vision
• 3D Geometry Processing
• Image Compression

Background:

• Accurate 3D geometry representation is crucial for various applications.
• Existing methods often result in large file sizes, hindering efficient storage and transmission.
• Integrating 3D data with 2D imagery presents a significant challenge.

Purpose of the Study:

• To develop a novel method for encoding 3D range geometry into a 2D RGB image.
• To enable high compression ratios for 3D data while maintaining reconstruction accuracy.
• To demonstrate the simultaneous encoding of 3D geometry and color texture within a single 2D image.

Main Methods:

• A two-frequency phase-shifting approach is utilized to encode 3D range geometry into two channels of a 2D RGB image.
• Conventional lossless or lossy image compression techniques are applied to the encoded 2D image.
• Experimental validation is performed to assess compression ratios and reconstruction accuracy.

Main Results:

• Achieved a compression ratio of 2883:1 compared to the STL format.
• Demonstrated high reconstruction accuracy with a Root Mean Square (RMS) error of 0.45 mm (99.8% accuracy) using JPEG 85 compression.
• Successfully encoded a 24-bit color texture map alongside 3D geometry within a single 2D image.

Conclusions:

• The proposed two-channel encoding method effectively integrates 3D geometry into 2D images.
• Significant data compression is achievable with high fidelity using standard image codecs.
• This approach offers a promising solution for efficient storage and transmission of 3D data with associated textures.