System for performing change detection within a 3D geospatial model based upon semantic change detection using deep learning and related methods
A system may include a memory and a processor cooperating therewith to obtain geospatial image data from a plurality of different types of sensors and generate a three-dimensional (3D) geospatial model therefrom. The processor may further determine a reference image within the 3D geospatial model based upon synthetically positioning an image sensor within the 3D geospatial model, and perform change detection between a collected image and the reference image based upon semantic change detection using deep learning.
1. A system comprising:
a memory and a processor cooperating therewith to
obtain geospatial image data from a plurality of different types of sensors and generate a three-dimensional (3D) geospatial model therefrom,
determine a reference image within the 3D geospatial model based upon synthetically positioning an image sensor within the 3D geospatial model, and
perform change detection between a collected image and the reference image based upon semantic change detection using deep learning.
2. The system of claim 1 wherein the processor is configured to perform the change detection based upon 3D elevational model differences.
3. The system of claim 1 wherein the processor is configured to perform the change detection based upon two-dimensional (2D) multi-modal imagery content changes.
4. The system of claim 1 wherein the processor is configured to perform the change detection based upon voxel-level historical volumetric attribution.
5. The system of claim 1 wherein the image sensor is synthetically positioned within the 3D geospatial model based upon locations of the plurality of different types of sensors during image capture.
6. The system of claim 1 wherein the processor is configured to perform change detection between the collected image and the reference image based upon semantic change detection using a convolutional neural network architecture (CNN).
7. The system of claim 1 wherein the processor is configured to simulate an atmospheric phenomena within the 3D geospatial model corresponding to an atmospheric phenomena at a time of image capture by at least one of the different types of sensors and prior to determining the reference image.
8. A system comprising:
a memory and a processor cooperating therewith to
obtain geospatial image data from a plurality of different types of sensors and generate a three-dimensional (3D) geospatial model therefrom,
simulate an atmospheric phenomena within the 3D geospatial model corresponding to an atmospheric phenomena at a time of image capture by at least one of the different types of sensors,
determine a reference image within the 3D geospatial model based upon synthetically positioning an image sensor within the 3D geospatial model and the simulated atmospheric phenomena, and
perform change detection between a collected image and the reference image based upon semantic change detection using a convolutional neural network architecture (CNN).
9. The system of claim 8 wherein the processor is configured to perform the change detection based upon at least one of 3D elevational model differences, two-dimensional (2D) multi-modal imagery content changes, and voxel-level historical volumetric attribution.
10. The system of claim 8 wherein the image sensor is synthetically positioned within the 3D geospatial model based upon locations of the different types of sensors during image capture.
11. A method comprising:
obtaining and storing geospatial image data in a memory from a plurality of different types of sensors; and
using a processor for
generating a three-dimensional (3D) geospatial model from the stored geospatial image data,
determining a reference image within the 3D geospatial model based upon synthetically positioning an image sensor within the 3D geospatial model, and
performing change detection between a collected image and the reference image based upon semantic change detection using deep learning.
12. The method of claim 11 wherein performing comprises performing the change detection based upon 3D elevational model differences.
13. The method of claim 11 wherein performing comprises performing the change detection based upon two-dimensional (2D) multi-modal imagery content changes.
14. The method of claim 11 wherein performing comprises performing the change detection based upon voxel-level historical volumetric attribution.
15. The method of claim 11 wherein the image sensor is synthetically positioned within the 3D geospatial model based upon locations of the plurality of different types of sensors during image capture.
16. The method of claim 11 wherein performing comprises performing change detection between the collected image and the reference image based upon semantic change detection using a convolutional neural network architecture (CNN).
17. The method of claim 11 comprising simulating, using the processor, an atmospheric phenomena within the 3D geospatial model corresponding to an atmospheric phenomena at a time of image capture by at least one of the different types of sensors and prior to determining the reference image.
18. A non-transitory computer-readable medium having computer-executable instructions for causing a processor to perform steps comprising:
obtaining and storing geospatial image data in a memory from a plurality of different types of sensors;
generating a three-dimensional (3D) geospatial model from the stored geospatial image data;
determining a reference image within the 3D geospatial model based upon synthetically positioning an image sensor within the 3D geospatial model; and
performing change detection between a collected image and the reference image based upon semantic change detection using deep learning.
19. The non-transitory computer-readable medium of claim 18 wherein performing comprises performing the change detection based upon at least one of 3D elevational model differences, two-dimensional (2D) multi-modal imagery content changes, and voxel-level historical volumetric attribution.
20. The non-transitory computer-readable medium of claim 18 wherein the image sensor is synthetically positioned within the 3D geospatial model based upon locations of the plurality of different types of sensors during image capture.
21. The non-transitory computer-readable medium of claim 18 wherein performing comprises performing change detection between the collected image and the reference image based upon semantic change detection using a convolutional neural network architecture (CNN).
22. The non-transitory computer-readable medium of claim 18 further having computer-executable instructions for causing the processor to perform a step of simulating an atmospheric phenomena within the 3D geospatial model corresponding to an atmospheric phenomena at a time of image capture by at least one of the different types of sensors and prior to determining the reference image.