Sparse Representation based Video Quality Assessment for Synthesized 3D Videos

Area of Science:

• Computer Vision
• Signal Processing
• Multimedia Systems

Background:

• Temporal flicker distortion is a significant artifact in synthesized virtual view videos within 3D systems.
• Accurate objective video quality assessment is crucial for optimizing compression techniques in 3D video rendering.

Purpose of the Study:

• To develop a full-reference objective video quality assessment method for synthesized 3D videos.
• To specifically address and measure temporal flicker distortion in multi-view video plus depth (MVD+) systems.

Main Methods:

• Reforming synthesized video as 3D volume data and decomposing into temporal layers (X-T, Y-T planes).
• Utilizing gradient features and depth map edges for flicker distortion localization.
• Applying dictionary learning and sparse representation to model temporal flicker.
• Employing a rank pooling method to aggregate temporal layer scores for flicker assessment.

Main Results:

• The proposed method effectively detects and quantifies temporal flicker distortion.
• Experimental results show superior performance compared to existing state-of-the-art methods.
• The method is particularly effective for view synthesis distortions originating from depth videos.

Conclusions:

• The developed sparse representation-based method provides accurate objective quality assessment for synthesized 3D videos.
• Integrating temporal flicker measurement with spatial distortion assessment enhances overall 3D video quality evaluation.
• This approach contributes to the optimization of compression and rendering in 3D video systems.