Reconstructing dense light field from array of multifocus images for novel view synthesis

Area of Science:

• Computer Vision
• Image Processing
• Computational Photography

Background:

• Synthesizing novel views requires dense light field data.
• Existing methods struggle with scenes containing multiple depth layers.
• Capturing all-in-focus images directly is challenging.

Purpose of the Study:

• To propose a novel method for synthesizing novel views from limited, differently focused images.
• To reconstruct a dense light field from sparse input data.
• To improve the quality of novel view synthesis for scenes with constant depths.

Main Methods:

• A two-step approach involving view interpolation and light-field rendering.
• View interpolation using linear filters for separate object region shifting without segmentation.
• Reconstruction of an all-in-focus dense light field.

Main Results:

• Successfully reconstructed a dense light field from two sets of differently focused images.
• Enabled synthesis of novel views with improved quality.
• Demonstrated effective handling of scenes with two approximate constant depths.

Conclusions:

• The proposed method effectively generates an all-in-focus dense light field.
• Novel view synthesis quality is significantly enhanced.
• The method offers a robust solution for specific scene types without complex region segmentation.