Redundancy of stereoscopic images: Experimental evaluation

Area of Science:

• Computer vision
• Image processing
• Human-computer interaction

Background:

• The growing market for stereoscopic content necessitates efficient transmission and display of 3D video.
• Transmitting stereoscopic content requires at least two viewpoints, impacting resource demands and system complexity.
• Stereoscopic images possess inherent redundancy, potentially beneficial for compression and visualization device design.

Purpose of the Study:

• To experimentally evaluate data redundancy in color stereoscopic images.
• To assess the impact of image blur on stereopsis and parallax measurement accuracy.
• To determine the color saturation limits for maintaining full 3D perception without visible color degradation.

Main Methods:

• Conducted experiments using computer-generated and real-life stereoscopic test images.
• Employed human observers to visually test stereopsis and parallax measurement thresholds.
• Varied the degree of blur in one of the stereo images.
• Investigated color saturation thresholds for maintaining 3D perception.

Main Results:

• Experimental results support theoretical estimates of data redundancy in stereoscopic images.
• A small percentage of additional data, beyond that of a single image, is sufficient for stereoscopic perception.
• Blurring one image impacts stereopsis and parallax accuracy, with specific thresholds observed.
• Full color 3D perception can be maintained within certain color saturation limits.

Conclusions:

• Significant data redundancy exists in stereoscopic images, offering opportunities for compression.
• Efficient stereoscopic content delivery is achievable with minimal additional data.
• Understanding blur and color saturation thresholds is crucial for optimizing stereoscopic display systems.