Aliasing Detection and Reduction Scheme on Angularly Undersampled Light Fields.

Area of Science:

• Computer Vision
• Image Processing
• Computational Photography

Background:

• Plenoptic cameras capture light fields for digital refocusing.
• Angular undersampling in light fields leads to severe aliasing artifacts.
• Prior methods focus on pre-processing to prevent aliasing.

Purpose of the Study:

• To present a novel approach for detecting and removing angular aliasing at the light field refocusing stage.
• To introduce a spatial domain analysis for aliasing detection and removal.
• To demonstrate advantages over existing light field processing techniques.

Main Methods:

• Developed a spatial domain analysis for identifying angular aliasing occurrence and location.
• Implemented a method to separate aliasing and non-aliasing regions.
• Applied the technique during the light field refocusing process.

Main Results:

• The spatial analysis effectively detects and locates angular aliasing artifacts.
• The proposed method successfully removes aliasing in both synthetic and real light field data.
• Demonstrated superior performance compared to prefiltering and depth-dependent rendering.

Conclusions:

• The presented spatial domain analysis offers an effective solution for angular aliasing in light field refocusing.
• This approach provides a significant improvement over traditional methods for digital refocusing.
• The technique is validated on diverse light field datasets, including camera array and Lytro data.