IMAGE BLENDING USING MULTI-SPLINES
First Claim
Patent Images
1. A computer-implemented process for blending image regions, comprising using a computer to perform the following process actions:
- associating a separate offset map with each image region being blended;
representing each offset map using a separate low-dimensional spline;
establishing an energy function comprising the summation of a set of terms comprising a separate term defining the difference in blended values of spline control points for each seam pixel and a separate term for each image region defining the blended values of the control points for each region, wherein a seam pixel is a pixel whose location corresponds to the boundary between two adjacent image regions;
solving the system of equation for the spline control point values that minimize the sum of all the terms;
computing a correction term for each pixel in each image region based on the spline control point values exhibited at spline control points and a prescribed 2D tensor product spline basis; and
applying the correction term associated with each pixel to that pixel to produce a blended image from the image regions.
2 Assignments
0 Petitions
Accused Products
Abstract
Multi-spline image blending technique embodiments are presented which generally employ a separate low-resolution offset field for every image region being blended, rather than a single (piecewise smooth) offset field for all the regions to produce a visually consistent blended image. Each of the individual offset fields is smoothly varying, and so is represented using a low-dimensional spline. A resulting linear system can be rapidly solved because it involves many fewer variables than the number of pixels being blended.
-
Citations
20 Claims
-
1. A computer-implemented process for blending image regions, comprising using a computer to perform the following process actions:
-
associating a separate offset map with each image region being blended; representing each offset map using a separate low-dimensional spline; establishing an energy function comprising the summation of a set of terms comprising a separate term defining the difference in blended values of spline control points for each seam pixel and a separate term for each image region defining the blended values of the control points for each region, wherein a seam pixel is a pixel whose location corresponds to the boundary between two adjacent image regions; solving the system of equation for the spline control point values that minimize the sum of all the terms; computing a correction term for each pixel in each image region based on the spline control point values exhibited at spline control points and a prescribed 2D tensor product spline basis; and applying the correction term associated with each pixel to that pixel to produce a blended image from the image regions. - View Dependent Claims (2, 3, 4)
-
-
5. A computer-implemented process for blending image regions, comprising using a computer to perform the following process actions:
-
respectively superimposing onto each image region a spline grid comprising spline control points having a prescribed spacing; identifying all seam pixels among the image regions, wherein a seam pixel is a pixel whose location corresponds to the boundary between two adjacent regions; for each seam pixel, computing a horizontal offset value representing the average of the intensity gradients between two adjacent image regions one of which is in the positive horizontal direction from the seam pixel under consideration, computing a vertical offset value represents the average of the intensity gradients between two adjacent image regions one of which is in the positive vertical direction from the seam pixel under consideration, computing a horizontal seam matching cost for the seam pixel under consideration which decreases as the horizontal offset value increases, computing a vertical seam matching cost for the seam pixel under consideration which decreases as the vertical offset value increases; computing a smoothness cost for each image region based on the variations in intensity between locations in the image region under consideration corresponding to a prescribed number of spline control points; establishing a seam pixel zero decay cost for each seam pixel and an image region zero decay cost for each image region; generating an energy function comprising, for each seam pixel, generating a term wherein a horizontally seam matched intensity term, a vertically seam matched intensity term and a seam pixel zero decay intensity term are summed, and wherein the seam pixel terms are dependent on unknown spline control point intensity values associated with the image regions forming the seam that the seam pixel under consideration belongs to, and for each image region, generating a term wherein a horizontally smoothed intensity term, a vertically smoothed intensity term and an image region zero decay intensity term are summed, and wherein the image region terms are dependent on unknown spline control point values associated with the image region under consideration; solving the system of equation for the spline control point values that minimize the sum of all the equations; computing a correction term for each pixel in each image region based on the spline control point values exhibited at spline control points and a prescribed 2D tensor product spline basis; and applying the correction term associated with each pixel to that pixel to produce a blended image from the image regions. - View Dependent Claims (6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
-
-
20. A computer-implemented process for blending image regions, comprising using a computer to perform the following process actions:
-
respectively superimposing onto each image region a spline grid comprising spline control points having a prescribed spacing S; identifying all seam pixels among the image regions, wherein a seam pixel is a pixel whose location corresponds to the boundary between two adjacent regions; for each seam pixel, computing a horizontal offset value gi,j−
x representing the average of the intensity gradients between two adjacent image regions one of which is in the positive horizontal direction from the seam pixel under consideration, where i,j refer to coordinates of the seam pixel,computing a vertical offset value gi,j−
y represents the average of the intensity gradients between two adjacent image regions one of which is in the positive vertical direction from the seam pixel under consideration,computing a horizontal seam matching cost si,jx for the seam pixel under consideration which decreases as the horizontal offset value increases, computing a vertical seam matching cost si,jy for the seam pixel under consideration which decreases as the vertical offset value increases; computing a smoothness cost s for each image region based on the variations in intensity between locations in the image region under consideration corresponding to a prescribed number of spline control points; establishing a seam pixel zero decay cost wi,j for each seam pixel and an image region zero decay cost w for each image region; generating a sum of squared terms comprising, a separate term for each seam pixel in the form of, where B(i+1−
kS,j−
mS), B(i−
kS,j−
mS) and B(i−
kS,j+1−
mS) represent various values of a 2D tensor product spline basis, k and m refer to the horizontal and vertical coordinates of the spline control points, ck,mli,j is the intensity exhibited at a spline control point having coordinates k,m if the point is assumed part of a first image region associated with the pixel at coordinates i,j, ck,mli+1,j is the intensity exhibited at a spline control point having coordinates k,m if the point is assumed part of an image region laying immediately adjacent the first image region in the positive horizontal direction, ck,mli,j+1 is the intensity exhibited at a spline control point having coordinates k,m if the point is assumed part of an image region laying immediately adjacent the first image region in the positive vertical direction, anda single term for each image region in the form of, where ck,m is the intensity exhibited by a spline control point overlying the image region under consideration and having coordinates k,m, ck+1,m is the intensity exhibited by a spline control point overlying the image region under consideration and having coordinates k+1,m, and ck,m+1 is the intensity exhibited by a spline control point overlying the image region under consideration and having coordinates k,m+1; solving the system of equation for the spline control point values that minimize the sum of all these terms; computing a correction term hi,jl for each pixel ij in each image region l based on the spline control point values ck,ml exhibited at spline control points having coordinates k,m and the a 2D tensor product spline basis B(i−
kS,j−
mS) using the equationapplying the correction term associated with each pixel to that pixel to produce a blended image from the image regions.
-
Specification