3D-MAD: A Full Reference Stereoscopic Image Quality Estimator Based on Binocular Lightness and Contrast Perception

Area of Science:

• Computer Vision
• Image Processing
• Human-Computer Interaction

Background:

• Stereoscopic image quality assessment (IQA) algorithms aim to predict human perception of 3D image quality.
• Current methods often adapt 2D IQA techniques to stereoscopic data, including disparity maps and cyclopean images.

Purpose of the Study:

• To introduce a novel 3D image quality assessment algorithm, named 3D-MAD.
• To enhance the accuracy of stereoscopic image quality evaluation by extending a 2D algorithm.

Main Methods:

• The 3D-MAD algorithm processes stereoscopic images in two stages: monocular view distortion and cyclopean view distortion.
• Stage 1 applies the 2D Most Apparent Distortion (MAD) algorithm to individual views, combining results using a contrast-based weighted sum.
• Stage 2 models the cyclopean view using a multipathway contrast gain-control model and statistical difference features.

Main Results:

• The algorithm achieved significant improvements over existing state-of-the-art 2D and 3D IQA algorithms on multiple test databases.
• The two-stage approach effectively captures quality degradation from both binocular and cyclopean perspectives.
• The weighted sum in Stage 1 and feature extraction in Stage 2 contribute to accurate quality prediction.

Conclusions:

• The proposed 3D-MAD algorithm offers a robust and accurate method for stereoscopic image quality assessment.
• It demonstrates superior performance compared to existing IQA algorithms, aligning better with human judgment.
• The algorithm provides a valuable tool for researchers and developers in the field of 3D imaging.