DIBR-Synthesized Image Quality Assessment With Texture and Depth Information

Area of Science:

• Computer Vision
• Image Processing
• Multimedia Engineering

Background:

• Depth-image-based-rendering (DIBR) synthesizes images from depth data.
• Existing image quality assessment (IQA) methods for DIBR images often overlook depth structure degradation.
• Accurate quality prediction is vital for advancing DIBR technologies.

Purpose of the Study:

• To develop a novel image quality assessment metric for DIBR-synthesized images.
• To address the limitations of current IQA methods that ignore depth structure damage.
• To improve the accuracy of visual quality evaluation for DIBR content.

Main Methods:

• Proposed a DIBR-synthesized image quality assessment metric named TDI (Texture and Depth Information).
• TDI jointly measures colorfulness, texture structure, and depth structure of synthesized images.
• The metric leverages the impact of DIBR on color deviation and local geometric distortions.

Main Results:

• TDI effectively quantifies distortions in DIBR-synthesized images by considering both visual and structural information.
• Experimental results demonstrate that TDI outperforms existing state-of-the-art IQA algorithms.
• The joint representation of texture and depth structures proves effective for accurate quality assessment.

Conclusions:

• The proposed TDI metric provides a more accurate evaluation of DIBR-synthesized image quality compared to existing methods.
• Considering colorfulness, texture, and depth structures is essential for robust DIBR IQA.
• TDI offers a significant advancement in assessing the visual fidelity of DIBR content.