Reconstruction of de-interleaved views, using adaptive interpolation based on disparity between the views for up-sampling
Method for reconstructing interleaved views of a 3D image comprising the steps of: receiving interleaved views of a 3D image, de-interleaving said interleaved views into a first view and a second view, each de-interleaved view comprising missing pixels, at each pixel location determining an amount of disparity between said first and second views, for each view interpolating each missing pixel based on the amount of disparity determined at the associated pixel location, wherein when the determined disparity is above a predetermined disparity threshold, the missing pixel value is interpolated from the data of the view to be interpolated only (i.e. the view comprising said missing pixel value), and when the determined disparity is at or below said predetermined disparity threshold, the missing pixel value is interpolated from the data of both views.
1. A method, comprising the steps of:
receiving a data stream containing interleaved left and right channel views of a 3D video stream;
de-interleaving the 3D video stream to extract the left and right channel views;
up-sampling the left and right channel views, comprising the steps of:
determining an amount of disparity between the left and right channel views at a pixel location or in a region corresponding to the pixel location; and
selecting at least one filter from a set of filters to interpolate a missing pixel value at the pixel location in one of the left and right channel views, wherein the filters from the set of filters correspond to different interpolation schemes,
wherein the at least one filter is selected based at least on the determined disparity such that,
when the determined disparity is at or below a disparity threshold, the at least one selected filter interpolates the missing pixel value either from data of both views or from data of only the other view, and
when the determined disparity is above the disparity threshold, the at least one selected filter interpolates the missing pixel value from data of only the view to be interpolated.
2. The method according to claim 1 ,
wherein, when the determined disparity is at or below the disparity threshold, the at least one selected filter interpolates the missing pixel value from data of only the other view.
3. The method according to claim 1 , wherein the data comprises temporal data from the view to be interpolated.
4. The method according to claim 1 , wherein one or more of the selected filters are Motion Compensated Temporal Filters (MCTF).
5. The method according to claim 1 , wherein the set of filters comprises a filter which considers multiple neighboring pixels.
6. The method according to claim 5 , wherein the at least one filter that is selected when the determined disparity is at or below the disparity threshold is a FIR filter applied on the neighboring pixels.
7. The method according to claim 1 , wherein the at least one filter that is selected when the determined disparity is at or below the disparity threshold is an iterative multi-stage filter, comprising an initial guess based on a co-located pixel in the other view determined by temporal interpolation based on at least one of motion adaptive and motion compensated methods.
8. The method according to claim 1 , wherein at least one of the filters comprises a pixel neighborhood that is specifically designed to address a deficiency caused by at least one of an interleaving scheme of the views in the data stream and a codec used on the video stream.
9. A device configured to up-sample different left and right views of an image and comprising a content adaptive interpolator,
wherein the content adaptive interpolator is configured to select at least one interpolation scheme to be utilized to up-sample at least a first view of the image based on an analysis of the first view and a second view of the image,
wherein the content adaptive interpolator comprises a disparity analysis module utilized to determine an interpolation scheme from a plurality of different interpolation schemes for interpolating a missing pixel value at a pixel location based at least on a determined amount of disparity between the first view and the second view at the pixel location or in a region corresponding to the pixel location such that,
when the determined disparity is at or below a disparity threshold, the content adaptive interpolator determines an interpolation scheme that interpolates the missing pixel value in the first view either from data of both views or from data of only the second view, and
when the determined disparity is above the disparity threshold, the content adaptive interpolator determines an interpolation scheme that interpolates the missing pixel value in the first view from data of only the first view.