Efficient Depth Enhancement Using a Combination of Color and Depth Information

Area of Science:

• Computer Vision
• Image Processing
• 3D Imaging

Background:

• Depth images are crucial for 3D applications but suffer from missing values and noisy boundaries.
• These imperfections degrade the performance of systems relying on depth data.
• Existing methods often struggle with accuracy and computational efficiency.

Purpose of the Study:

• To develop an effective depth enhancement algorithm.
• To address issues of missing depth values and noisy object boundaries.
• To improve the overall quality of depth images for various applications.

Main Methods:

• A novel algorithm combining color and depth information for depth enhancement.
• Utilizing asynchronous cellular automata with neighborhood distance maps to fill depth holes and recover shapes.
• Employing image segmentation and weighted linear spatial filtering to refine object regions and handle disocclusions.

Main Results:

• The proposed method significantly enhances depth image quality.
• Demonstrated low computational complexity compared to conventional techniques.
• Outperformed existing methods on multiple performance metrics across real-world and public datasets.

Conclusions:

• The developed depth enhancement algorithm effectively improves depth image quality.
• The method offers a computationally efficient solution for 3D data processing.
• Enhanced depth images show improved quality, verified by generated stereoscopic images.