Techniques for spatial displacement estimation and multi-resolution operations on light fields

A method for selectively accessing enhancement information for a set of plural spatially-related views of an object or static scene, wherein a data structure stores a spatially-related view, the data structure comprising a base field and an enhancement field, the method comprising:

• for a point of a novel perspective view, determining at least one match point in the view stored in the data structure;
• for a match point of the at least one match point in the view,
• accessing information in the base field;
• and
• selectively accessing information in the enhancement field.

A method for compressing a representation of a set of plural spatially-related views by reducing spatial redundancy between the spatially-related views, the method comprising:

• for a prediction view from a set of plural spatially-related views,
• for each of plural sections of the prediction view,
• determining a corresponding section predictor in a reference view;
• and
• determining a displacement vector based upon the corresponding section predictor, wherein the displacement vector indicates a spatial transformation;
• and
• representing the prediction view based upon displacement vectors determined for the plural sections of the prediction view.

A method for rendering a novel perspective view from a set of plural compressed spatially-related views of an object or static scene, the method comprising:

• decompressing information for one or more reference views of the set of spatially-related views;
• decompressing information for a prediction view of the set of spatially-related views, wherein the prediction view estimates spatial displacement from at least one reference view;
• for a point of a novel perspective view,
• determining at least one match point in the decompressed views;
• calculating a reproduction value based upon decompressed information for the at least one match points;
• and
• assigning the reproduction value to the point of the novel perspective view.

A method for configuring a set of spatially-related views, the method comprising:

• arranging a set of spatially-related views in a configuration that reflects the geometrical relationships between the spatially-related views;
• designating as reference views one or more views from the set of spatially-related views;
• designating as prediction views one or more views from the set of spatially-related views that are not designated as reference views;
• representing each prediction view in terms of plural displacement vectors from one or more of the reference views.