Augmented reality processing device and method

1. An augmented reality processing device, comprising:

an image capturing circuit, configured to capture a first environment image and a second environment image; and

a processor, connected to the image capturing circuit, and execute operations of:

generating an original point cloud image according to the first environment image and a physical object in the first environment image, wherein the original point cloud image comprises a physical object point cloud set and a physical scene point cloud set which correspond to the physical object, wherein a virtual object is superimposed with the physical object according to the physical object point cloud set;

generating a first physical object feature set corresponding to the physical object point cloud set according to the first environment image and the physical object point cloud set;

generating an expanded point cloud image corresponding to the physical object from the second environment image according to the first physical object feature set, and generating a superimposed point cloud image according to the expanded point cloud image and the original point cloud image; and

generating a transformation matrix according to the original point cloud image and the expanded point cloud image, and superimposing the virtual object to the second environment image according to the superimposed point cloud image and the transformation matrix.

2. The augmented reality processing device of claim 1 , wherein the processor is further configured for:

executing ORB algorithms according to the first environment image and the physical object point cloud set to generate the first physical object feature set corresponding to the physical object point cloud set.

3. The augmented reality processing device of claim 1 , wherein the processor is further configured for:

identifying an object area of the physical object in the second environment image according to the first physical object feature set, and generating the expanded point cloud image corresponding to the physical object in the second environment image from the object area.

4. The augmented reality processing device of claim 3 , wherein the processor is further configured for:

superimposing the expanded point cloud image to a position corresponding to the object area in the original point cloud image to generate the superimposed point cloud image.

5. The augmented reality processing device of claim 3 , wherein the processor is further configured for:

performing ORB algorithms on the object area in the second environment image to generate a second physical object feature set corresponding to the physical object.

6. The augmented reality processing device of claim 5 , wherein the processor is further configured for:

establishing a point cloud correspondence relationship between the physical object point cloud set and the expanded point cloud image according to the first physical object feature set and the second physical object feature set; and

generating the transformation matrix according to the point cloud correspondence relationship, the physical object point cloud set, and the expanded point cloud image.

7. The augmented reality processing device of claim 6 , wherein the processor is further configured for:

utilizing the transformation matrix to execute geometric transformation on the expanded point cloud image, so as to generate an adjusted expanded point cloud image, and executing distance function processing according to the adjusted expanded point cloud image and the expanded point cloud image to generate a distance value; and

determining, according to the distance value and a distance threshold, whether to generate an adjusted transformation matrix according to the point cloud correspondence relationship, the physical object point cloud set, and the adjusted expanded point cloud image, so as to superimpose the virtual object on the second environment image according to the superimposed point cloud image and the adjusted transformation matrix.

8. The augmented reality processing device of claim 6 , wherein the processor is further configured for:

calculating a barycenter position corresponding to the physical object point cloud set and a barycenter position corresponding to the expanded point cloud image according to the physical object point cloud set and the expanded point cloud image; and

generating the transformation matrix according to the point cloud correspondence relationship, the barycenter position corresponding to the physical object point cloud set, and the barycenter position corresponding to the expanded point cloud image.

9. The augmented reality processing device of claim 1 , wherein the processor is further configured for:

utilizing the transformation matrix to execute geometric transformation on the virtual object, so as to generate a transformed virtual object, and superimposing the transformed virtual object on the second environment image according to the superimposed point cloud image.

10. The augmented reality processing device of claim 1 , wherein the first environment image and the second environment image respectively correspond to a first capturing time and a second capturing time, wherein the first capturing time is before the second capturing time.

11. An augmented reality processing method, comprising:

capturing a first environment image and a second environment image, and generating an original point cloud image according to the first environment image and a physical object in the first environment image, wherein the original point cloud image comprises a physical object point cloud set and a physical scene point cloud set which correspond to the physical object, wherein a virtual object is superimposed with the physical object according to the physical object point cloud set;

generating a first physical object feature set corresponding to the physical object point cloud set according to the first environment image and the physical object point cloud set;

generating an expanded point cloud image corresponding to the physical object from the second environment image according to the first physical object feature set, and generating a superimposed point cloud image according to the expanded point cloud image and the original point cloud image; and

generating a transformation matrix according to the original point cloud image and the expanded point cloud image, and superimposing the virtual object to the second environment image according to the superimposed point cloud image and the transformation matrix.

12. The augmented reality processing method of claim 11 , wherein the step of generating the first physical object feature set corresponding to the physical object point cloud set according to the first environment image and the physical object point cloud set comprises:

executing ORB algorithms according to the first environment image and the physical object point cloud set to generate the first physical object feature set corresponding to the physical object point cloud set.

13. The augmented reality processing method of claim 11 , wherein the step of generating the expanded point cloud image corresponding to the physical object from the second environment image according to the first physical object feature set comprises:

identifying an object area of the physical object in the second environment image according to the first physical object feature set, and generating the expanded point cloud image corresponding to the physical object in the second environment image from the object area.

14. The augmented reality processing method of claim 13 , wherein the step of generating the superimposed point cloud image according to the expanded point cloud image and the original point cloud image comprises:

superimposing the expanded point cloud image to a position corresponding to the object area in the original point cloud image to generate the superimposed point cloud image.

15. The augmented reality processing method of claim 13 , further comprising:

performing ORB algorithms on the object area in the second environment image to generate a second physical object feature set corresponding to the physical object.

16. The augmented reality processing method of claim 15 , wherein the step of generating the transformation matrix according to the original point cloud image and the expanded point cloud image comprises:

establishing a point cloud correspondence relationship between the physical object point cloud set and the expanded point cloud image according to the first physical object feature set and the second physical object feature set; and

generating the transformation matrix according to the point cloud correspondence relationship, the physical object point cloud set, and the expanded point cloud image.

17. The augmented reality processing method of claim 16 , wherein the step of generating the transformation matrix according to the original point cloud image and the expanded point cloud image comprises:

utilizing the transformation matrix to execute geometric transformation on the expanded point cloud image, so as to generate an adjusted expanded point cloud image, and executing distance function processing according to the adjusted expanded point cloud image and the expanded point cloud image to generate a distance value; and

determining, according to the distance value and a distance threshold, whether to generate an adjusted transformation matrix according to the point cloud correspondence relationship, the physical object point cloud set, and the adjusted expanded point cloud image, so as to superimpose the virtual object on the second environment image according to the superimposed point cloud image and the adjusted transformation matrix.

18. The augmented reality processing method of claim 16 , wherein the step of generating the transformation matrix according to the original point cloud image and the expanded point cloud image comprises:

calculating a barycenter position corresponding to the physical object point cloud set and a barycenter position corresponding to the expanded point cloud image according to the physical object point cloud set and the expanded point cloud image; and

generating the transformation matrix according to the point cloud correspondence relationship, the barycenter position corresponding to the physical object point cloud set, and the barycenter position corresponding to the expanded point cloud image.

19. The augmented reality processing method of claim 11 , wherein the step of superimposing the virtual object to the second environment image according to the superimposed point cloud image and the transformation matrix comprises:

utilizing the transformation matrix to execute geometric transformation on the virtual object, so as to generate a transformed virtual object, and superimposing the transformed virtual object on the second environment image according to the superimposed point cloud image.

20. The augmented reality processing method of claim 11 , wherein the first environment image and the second environment image respectively correspond to a first capturing time and a second capturing time, wherein the first capturing time is before the second capturing time.