Method of rendering an output image on basis of an input image and a corresponding depth map

1. A method of rendering an output image on basis of an input image and a corresponding matrix of depth related data, the input image corresponding to a first viewpoint and the output image corresponding to a second viewpoint being different from the first viewpoint, the method comprising:

computing edges, via an edge detection unit, in the matrix of depth related data, by computing only derivatives of the depth related data in a predetermined direction (X), the edges corresponding to edges between (i) an object in a foreground of the input image and (ii) a background of the input image;

computing a measure of variation in pixel values of the input image, via a texture analyzing unit, in regions of the input image that form a pixel subset (S vp ) that depends on the second viewpoint corresponding to neighborhoods of selected ones of the edges in de-occlusion areas as a function of the second viewpoint, wherein the neighborhoods in the de-occlusion areas are located at a background side of the selected ones of the edges, and wherein the computed measure of variation is a texture measure (T vp ) that comprises an average horizontal color change in the input image from one pixel to the next in the predetermined direction (X), averaged over a corresponding region of the pixel subset (S vp ); and

computing the output image, via scaling and rendering units, by shifting respective pixel values of the input image in the predetermined direction (X) on basis of scaled depth related data, wherein the scaled depth related data comprises depth related data having been scaled on basis of the computed measure of variation, further wherein scaling via the scaled depth related data is such that a relative large measure of variation in pixel values in the regions of the input image that form the pixel subset (S vp ) of corresponding neighborhoods of the select ones of the edges in the de-occlusion areas results in a relative large reduction of depth perception via a reduction in dynamic range of depth values in the scaled depth related data scaled on the basis of the computed measure of variation.

2. The method of rendering an output image as claimed in claim 1 , wherein for computing, via the edge detection unit, the edges in the matrix of depth related data:

more positive derivatives of the depth related data are used than negative derivatives, if the second viewpoint is located on a positive side of the first viewpoint, in the predetermined direction (X); and

more negative derivatives of the depth related data are used than positive derivatives, if the second viewpoint is located on a negative side of the first viewpoint, in the predetermined direction (X).

3. The method of rendering an output image as claimed in claim 1 , wherein for computing, via the edge detection unit, the edges in the matrix of depth related data, the magnitudes of the respective derivatives are larger than a predetermined threshold.

4. The method of rendering an output image as claimed in claim 1 , wherein a first border of a particular one of the regions of the input image is determined on basis of the corresponding edge and wherein the opposite border is located at a predetermined number of pixels from the first border, in the predetermined direction (X).

5. The method of rendering an output image as claimed in claim 1 , wherein scaling is performed globally in the input image.

6. The method of rendering an output image as claimed in claim 1 , wherein scaling is performed locally in the input image depending on a locally computed measure of variation.

7. An image rendering unit for rendering an output image on basis of an input image and a corresponding matrix of depth related data, the input image corresponding to a first viewpoint and the output image corresponding to a second viewpoint being different from the first viewpoint, the image rendering unit comprising:

first computing means for computing edges in the matrix of depth related data, by computing only derivatives of the depth related data in a predetermined direction (X), the edges corresponding to edges between (i) an object in a foreground of the input image and (ii) a background of the input image;

second computing means for computing a measure of variation in pixel values of the input image in regions of the input image that form a pixel subset (S vp ), that depends on the second viewpoint corresponding to neighborhoods of selected ones of the edges in de-occlusion areas as a function of the second viewpoint, wherein the neighborhoods in the de-occlusion areas are located at a background side of the selected ones of the edges, and wherein the computed measure of variation is a texture measure (T vp ) that comprises an average horizontal color change in the input image from one pixel to the next in the predetermined direction (X), averaged over a corresponding region of the pixel subset (S vp ); and

third computing means for computing the output image by shifting respective pixel values of the input image in the predetermined direction (X) on basis of scaled depth related data, wherein the scaled depth related data comprises depth related data having been scaled on basis of the computed measure of variation, further wherein scaling via the scaled depth related data is such that a relative large measure of variation in pixel values in the regions of the input image that form the pixel subset (S vp ) of corresponding neighborhoods of the select ones of the edges in the de-occlusion areas results in a relative large reduction of depth perception via a reduction in dynamic range of depth values in the scaled depth related data scaled on the basis of the computed measure of variation.

8. An image processing apparatus comprising:

receiving means for receiving an input image and a corresponding matrix of depth related data; and

an image rendering unit for rendering an output image on basis of the input image and the corresponding matrix of depth related data, as claimed in claim 5 .

9. A non-transitory computer readable medium encoded with a computer program to be loaded by a computer arrangement, comprising instructions to compute an output image on basis of an input image and a corresponding matrix of depth related data, the input image corresponding to a first viewpoint and the output image corresponding to a second viewpoint being different from the first viewpoint, the computer arrangement comprising processing means and a memory, the computer program, after being loaded, providing said processing means with the capability to carry out:

computing edges in the matrix of depth related data, by computing only derivatives of the depth related data in a predetermined direction (X), the edges corresponding to edges between (i) an object in a foreground of the input image and (ii) a background of the input image;

computing a measure of variation in pixel values of the input image in regions of the input image that form a pixel subset (S vp ), that depends on the second viewpoint corresponding to neighborhoods of selected ones of the edges in de-occlusion areas as a function of the second viewpoint, wherein the neighborhoods in the de-occlusion areas are located at a background side of the selected ones of the edges, and wherein the computed measure of variation is a texture measure (T vp ) that comprises an average horizontal color change in the input image from one pixel to the next in the predetermined direction (X), averaged over a corresponding region of the pixel subset (S vp ); and

computing the output image by shifting respective pixel values of the input image in the predetermined direction (X) on basis of scaled depth related data, wherein the scaled depth related data comprises depth related data having been scaled on basis of the computed measure of variation, further wherein scaling via the scaled depth related data is such that a relative large measure of variation in pixel values in the regions of the input image that form the pixel subset (S vp )of corresponding neighborhoods of the selected ones of the edges in the de-occlusion areas results in a relative large reduction of depth perception via a reduction in dynamic range of depth values in the scaled depth related data scaled on the basis of the computed measure of variation.