Joint bilateral upsampling
View Patent ↗A “Joint Bilateral Upsampler” uses a high-resolution input signal to guide the interpolation of a low-resolution solution set (derived from a downsampled version of the input signal) from low-to high-resolution. The resulting high-resolution solution set is then saved or applied to the original input signal to produce a high-resolution output signal. The high-resolution solution set is close to what would be produced directly from the input signal without downsampling. However, since the high-resolution solution set is constructed in part from a downsampled version of the input signal, it is computed using significantly less computational overhead and memory than a solution set computed directly from a high-resolution signal. Consequently, the Joint Bilateral Upsampler is advantageous for use in near real-time operations, in applications where user wait times are important, and in systems where computational costs and available memory are limited.
1. A method for upsampling a solution set, comprising using a computing device to:
perform steps for:
receiving a high-resolution input signal having a first resolution;
receiving a low-resolution solution set computed from a downsampled version of the high-resolution signal, said low-resolution solution set having a second resolution lower than the first resolution;
performing a joint bilateral upsampling using the low-resolution solution set and the high-resolution signal to generate a high-resolution solution set having a resolution equivalent to the first resolution; and
storing the high resolution solution set for later use by a user.
2. The method of claim 1 wherein the input signal is an image.
3. The method claim 2 wherein the low-resolution solution set represents an exposure solution for tone mapping the input image.
4. The method claim 2 wherein the low-resolution solution set represents a pair of chromaticity solutions for colorizing the input image.
5. The method claim 2 wherein the low-resolution solution set represents a disparity solution computing depth map from two or more input images.
6. The method claim 2 wherein the low-resolution solution set represents a label map solution of a graph-cut optimization.
7. The method of claim 1 wherein the input signal is a 3D triangular mesh representing a 3D object model, and wherein the low-resolution solution set represents a global illumination model for the 3D object model.
8. The system of claim 1 further comprising processing the high-resolution input signal using the high-resolution solution set to produce a high-resolution output signal.
9. A physical computer-readable storage device having computer executable instructions stored therein for generating a high-resolution output signal from a high-resolution input signal, said computer executable instructions comprising:
receiving a high-resolution input signal;
downsampling the input signal to produce a downsampled signal;
processing the downsampled signal to generate a low-resolution signal solution;
performing a joint bilateral filtering of the low-resolution signal solution and the high-resolution input signal to generate a high-resolution signal solution; and
applying the high-resolution signal solution to the high-resolution input signal to generate a high resolution output signal.
10. The computer-readable storage device of claim 9 wherein the high-resolution input signal is at least one image.
11. The computer-readable storage device of claim 10 wherein the low-resolution signal solution represents an exposure solution for tone mapping the input image.
12. The computer-readable storage device of claim 10 wherein the low-resolution signal solution represents a pair of chromaticity solutions for colorizing the input image.
13. The computer-readable storage device of claim 10 wherein the low-resolution signal solution represents a disparity solution computing depth map from two or more input images.
14. The computer-readable storage device of claim 10 wherein the low-resolution signal solution represents a label map solution of a graph-cut optimization.
15. The computer-readable storage device of claim 10 wherein the input signal is a 3D triangular mesh representing a 3D object model, and wherein the low-resolution signal solution represents a global illumination model for the 3D object model.
16. A camera for processing images in near real-time, comprising using a computing device embedded within the camera for:
capturing a high-resolution input image;
downsampling the input image;
processing the downsampled input image to generate a low-resolution image solution;
performing a joint bilateral filtering of the low-resolution image solution relative to the high-resolution input image to generate a high-resolution image solution; and
applying the high-resolution image solution to the high-resolution input image to generate a high-resolution output image.
17. The camera of claim 16 wherein the low-resolution image solution represents an exposure solution for tone mapping the input image.
18. The camera of claim 16 wherein the low-resolution image solution represents a pair of chromaticity solutions for colorizing the input image.
19. The camera of claim 16 wherein the low-resolution image solution represents a disparity solution computing depth map from two or more input images.
20. The camera of claim 16 wherein the low-resolution image solution represents a label map solution of a graph-cut optimization.