Dynamic real-time texture alignment for 3D models

1. A system for dynamic real-time texture alignment for three-dimensional (3D) models, the system comprising a memory that stores computer-executable instructions and a processor that executes the computer-executable instructions to:

receive a plurality of input images of one or more objects in a scene;

generate a 3D model for at least one of the one or more objects in the scene, the 3D model comprising a plurality of mesh triangles;

project each mesh triangle of the 3D model to one of the plurality of input images;

measure a texture discontinuity between adjacent mesh triangles of the projected input image by comparing one or more color differences in a shared edge of the adjacent mesh triangles;

translate a texture associated with each of the adjacent mesh triangles in a plurality of different directions to create texture candidates for each of the adjacent mesh triangles;

apply each of the texture candidates to the corresponding mesh triangles until a seamless texture join is formed on the shared edge, comprising:

determining an optimal texture labeling for each mesh triangle of the projected input image using an energy minimization algorithm;

downsampling the projected input image to generate an intermediate image, the intermediate image comprising a decreased resolution in one or more dimensions;

determining an optimal texture labeling for each mesh triangle of the intermediate image using the energy minimization algorithm and the optimal texture labeling for each mesh triangle from the projected input image;

downsampling the intermediate image to generate a target image;

determining an optimal texture labeling for each mesh triangle of the target image using the energy minimization algorithm and the optimal texture labeling for each mesh triangle from the intermediate image;

selecting a texture candidate that corresponds to the optimal texture labeling for each mesh triangle of the target image; and

applying the selected texture candidate to the corresponding mesh triangle of the projected input image; and

generate a textured 3D model using the 3D model, the projected input image, and the texture candidates that form the seamless texture join for each adjacent mesh triangle.

2. The system of claim 1 , wherein the optimal texture labeling for each mesh triangle of the target image comprises a texture labeling associated with a minimum energy value determined by the energy minimization algorithm using the optimal texture labeling for each mesh triangle from the intermediate image as input.

3. The system of claim 1 , wherein the optimal texture labeling for each mesh triangle of the intermediate image comprises a texture labeling associated with a minimum energy value determined by the energy minimization algorithm using the optimal texture labeling for each mesh triangle from the projected input image as input.

4. The system of claim 1 , wherein the optimal texture labeling for each mesh triangle of the projected input image comprises a texture labeling associated with a minimum energy value determined by the energy minimization algorithm.

5. The system of claim 1 , wherein the energy minimization algorithm is an alpha-expansion algorithm.

6. The system of claim 1 , wherein the plurality of input images are high-definition (HD) images.

7. The system of claim 1 , wherein information corresponding to the textured 3D model is stored as an .obj file, an .mtl file, and a .jpg file.

8. The system of claim 1 , wherein translating the texture associated with each of the adjacent mesh triangles in a plurality of different directions comprises translating each texture in nine different directions.

9. A computerized method of dynamic real-time texture alignment for three-dimensional (3D) models, the method comprising:

receiving, by a computing device, a plurality of input images of one or more objects in a scene;

generating, by the computing device, a 3D model for at least one of the one or more objects in the scene, the 3D model comprising a plurality of mesh triangles;

projecting, by the server computing device, each mesh triangle of the 3D model to one of the plurality of input images;

measuring, by the server computing device, a texture discontinuity between adjacent mesh triangles of the projected input image by comparing one or more color differences in a shared edge of the adjacent mesh triangles;

translating, by the server computing device, a texture associated with each of the adjacent mesh triangles in a plurality of different directions to create texture candidates for each of the adjacent mesh triangles;

applying, by the server computing device, each of the texture candidates to the corresponding mesh triangles until a seamless texture join is formed on the shared edge, comprising:

determining an optimal texture labeling for each mesh triangle of the projected input image using an energy minimization algorithm;

downsampling the projected input image to generate an intermediate image, the intermediate image comprising a decreased resolution in one or more dimensions;

determining an optimal texture labeling for each mesh triangle of the intermediate image using the energy minimization algorithm and the optimal texture labeling for each mesh triangle from the projected input image;

downsampling the intermediate image to generate a target image;

determining an optimal texture labeling for each mesh triangle of the target image using the energy minimization algorithm and the optimal texture labeling for each mesh triangle from the intermediate image;

selecting a texture candidate that corresponds to the optimal texture labeling for each mesh triangle of the target image; and

applying the selected texture candidate to the corresponding mesh triangle of the projected input image; and

generating, by the server computing device, a textured 3D model using the 3D model, the projected input image, and the texture candidates that form the seamless texture join for each adjacent mesh triangle.

10. The method of claim 9 , wherein the optimal texture labeling for each mesh triangle of the target image comprises a texture labeling associated with a minimum energy value determined by the energy minimization algorithm using the optimal texture labeling for each mesh triangle from the intermediate image as input.

11. The method of claim 9 , wherein the optimal texture labeling for each mesh triangle of the intermediate image comprises a texture labeling associated with a minimum energy value determined by the energy minimization algorithm using the optimal texture labeling for each mesh triangle from the projected input image as input.

12. The method of claim 9 , wherein the optimal texture labeling for each mesh triangle of the projected input image comprises a texture labeling associated with a minimum energy value determined by the energy minimization algorithm.

13. The method of claim 9 , wherein the energy minimization algorithm is an alpha-expansion algorithm.

14. The method of claim 9 , wherein the plurality of input images are high-definition (HD) images.

15. The method of claim 9 , wherein information corresponding to the textured 3D model is stored as an .obj file, an .mtl file, and a .jpg file.

16. The method of claim 9 , wherein translating the texture associated with each of the adjacent mesh triangles in a plurality of different directions comprises translating each texture in nine different directions.