RDTC optimized compression of image-based scene representations (Part II): practical coding.

Area of Science:

• Computer Vision
• Image Processing
• Multimedia Systems

Background:

• Interactive streaming of image-based scene representations necessitates random access to reference image data.
• Existing methods often overlook the combined impact of transmission data rate and decoding complexity on compression and streaming optimization.
• The traditional rate-distortion (RD) tradeoff is insufficient for optimizing streaming performance.

Purpose of the Study:

• To extend the rate-distortion (RD) optimization to a rate-distortion-transmission-complexity (RDTC) tradeoff for image-based scene representations.
• To develop a practical procedure for modeling and selecting encoding parameters tailored to scenario-specific properties.
• To evaluate the impact of client-side caching on RDTC-optimized streams.

Main Methods:

• Theoretical analysis of the RDTC space for densely sampled image-based scene representations.
• Development of a practical RDTC optimization procedure for adaptive compression.
• Evaluation using an experimental testbed incorporating client-side caching.

Main Results:

• RDTC optimization enables adaptation of compression to specific scenario properties.
• Client-side caching was integrated and evaluated within the experimental testbed.
• RDTC-optimized streams demonstrated significant reductions in user-perceived delay, memory usage, and required bitrates compared to traditional methods.

Conclusions:

• Practical RDTC optimization offers a superior approach for compressing and streaming image-based scene representations.
• The proposed method effectively balances rate, distortion, transmission, and decoding complexity for enhanced user experience.
• This approach is crucial for efficient interactive streaming applications demanding low latency and reduced resource utilization.