Geometric method of transforming a two-dimensional image

1. A geometric method implemented by a processing circuit of transforming images, comprising:

dividing one of said first and second images into a plurality of surfaces;

applying, to each surface of the divided image, a bilinear transformation specific to each surface to obtain transformed surfaces; and

grouping the transformed surfaces together;

wherein said bilinear transformation approximates an inverse equirectangular transformation and a gnomonic transformation.

2. The method of claim 1 , further comprising choosing a bilinear transformation for each surface such that the vertices of the transformed surface are identical to vertices that an inverse equirectangular transformation and a gnomonic transformation would produce.

3. The method of claim 1 wherein:

dividing one of the first and second images comprises slicing the second image into quadrilateral surfaces; and

applying, to each surface of the divided image, a bilinear transformation specific to each surface to obtain transformed surfaces comprises:

determining four points in the first image corresponding to vertices of the quadrilateral surface; and

filling the quadrilateral surface through a bilinear transformation which associates said four vertices of the quadrilateral surface with said four points.

4. The method of claim 3 , wherein

said bilinear transformation fills the quadrilateral surface using a surface on the first image delimited by said four points.

5. The method of claim 3 , further comprising, before grouping the transformed surfaces together and for each of the quadrilateral surfaces:

determining the error between the quadrilateral surface after filling and a reference surface arising from an inverse equirectangular transformation and a gnomonic transformation on the whole of the surface of the first image delimited by said four points; and

if this error is greater than a predetermined threshold, subdividing the quadrilateral surface.

6. The method of claim 5 wherein determining the error between the quadrilateral surface after filling and the reference surface comprises:

determining a first point in the reference surface associated with the center of the quadrilateral surface through said bilinear transformation;

determining a second point in the reference surface associated with the center of the quadrilateral surface through an inverse equirectangular transformation and a gnomonic transformation; and

computing the distance between the first and second points.

7. The method of claim 3 wherein filling of each of the quadrilateral surfaces through the bilinear transformation is carried out in geometric form and comprises:

determining, by using the properties of conservation of the barycentres of a bilinear transformation, lines in the surface on the first image delimited by said four points that result from said bilinear transformation applied to lines of pixels in the quadrilateral surface;

determining, by using the properties of conservation of the barycentres of a bilinear transformation, locations in the surface on the first image delimited by said four points that result from said bilinear transformation applied to pixels in the quadrilateral surface; and

filling the pixels in the quadrilateral surface with the values of said locations in the surface on the first image delimited by said four points.

8. The method of claim 1 wherein:

dividing one of said first and second images into a plurality of surfaces comprises dividing the first image into a plurality of surfaces; and

applying, to each surface of the divided image, a bilinear transformation specific to each surface to obtain transformed surfaces comprises applying, to each surface of the first image, a bilinear transformation that maps points from the surface to a corresponding surface of the second image.

9. The method of claim 1 wherein:

dividing one of said first and second images into a plurality of surfaces comprises dividing the second image into a plurality of surfaces; and

applying, to each surface of the divided image, a bilinear transformation specific to each surface to obtain transformed surfaces comprises applying, to each surface of the second image, a bilinear transformation that maps points of the surface from a corresponding surface of the first image.

10. A geometric method implemented by a processing circuit of transforming images, comprising:

dividing one of said first and second images into a plurality of surfaces;

applying, to each surface of the divided image, a bilinear transformation specific to each surface to obtain transformed surfaces; and

grouping the transformed surfaces together;

wherein said bilinear transformation approximates a perspective transformation.

11. The method of claim 10 , further comprising choosing a bilinear transformation for each surface such that the vertices of the transformed surface are identical to vertices that a perspective transformation would produce.

12. The method of claim 10 , wherein:

dividing one of the first and second images comprises slicing the second image into quadrilateral surfaces; and

applying, to each surface of the divided image, a bilinear transformation specific to each surface to obtain transformed surfaces comprises:

determining four points in the first image corresponding to vertices of the quadrilateral surface; and

filling the quadrilateral surface through a bilinear transformation which associates said four vertices of the quadrilateral surface with said four points.

13. The method of claim 12 , wherein

said bilinear transformation fills the quadrilateral surface using a surface on the first image delimited by said four points.

14. The method of claim 12 , further comprising, before grouping the transformed surfaces together and for each of the quadrilateral surfaces:

determining the error between the quadrilateral surface after filling and a reference surface arising from a perspective transformation on the whole of the surface of the first image delimited by said four points; and

if this error is greater than a predetermined threshold, subdividing the quadrilateral surface.

15. The method of claim 14 wherein determining the error between the quadrilateral surface after filling and the reference surface comprises:

determining a first point in the reference surface associated with the center of the quadrilateral surface through said bilinear transformation;

determining a second point in the reference surface associated with the center of the quadrilateral surface through a perspective transformation; and

computing the distance between the first and second points.

16. The method of claim 12 wherein filling of each of the quadrilateral surfaces through the bilinear transformation is carried out in geometric form and comprises:

determining, by using the properties of conservation of the barycentres of a bilinear transformation, lines in the surface on the first image delimited by said four points that result from said bilinear transformation applied to lines of pixels in the quadrilateral surface;

determining, by using the properties of conservation of the barycentres of a bilinear transformation, locations in the surface on the first image delimited by said four points that result from said bilinear transformation applied to pixels in the quadrilateral surface; and

filling the pixels in the quadrilateral surface with the values of said locations in the surface on the first image delimited by said four points.

17. The method of claim 10 wherein:

dividing one of said first and second images into a plurality of surfaces comprises dividing the first image into a plurality of surfaces; and

applying, to each surface of the divided image, a bilinear transformation specific to each surface to obtain transformed surfaces comprises applying, to each surface of the first image, a bilinear transformation that maps points from the surface to a corresponding surface of the second image.

18. The method of claim 10 wherein:

dividing one of said first and second images into a plurality of surfaces comprises dividing the second image into a plurality of surfaces; and

applying, to each surface of the divided image, a bilinear transformation specific to each surface to obtain transformed surfaces comprises applying, to each surface of the second image, a bilinear transformation that maps points of the surface to a corresponding surface of the first image.