Neural radiance field ray construction for true nadir renderings
Extraction of depth from satellite images using neural radiance fields may be provided. Satellite image data may be received. Then a plurality of nadir rays, each with a plurality of points, in a first space may be constructed. A Machine Learning (ML) model may then be used to map the plurality of points along the plurality of nadir rays in the first space to a second space. Then the plurality of points along the plurality of nadir rays in the second space may be rendered into images and depths.
1 . A method comprising:
receiving satellite image data;
constructing a plurality of nadir rays, each of the nadir rays extending from a unique point of origin to a pixel from the image data, whereby the plurality of nadir rays are parallel to one another, each of the plurality of nadir rays further including a plurality of points, in a first space;
using a Machine Learning (ML) model to map the plurality of points along the plurality of nadir rays in the first space to a second space; and
rendering the plurality of points along the plurality of nadir rays in the second space into images.
2 . The method of claim 1 , wherein the ML model comprises a Multilayer Perceptron (MLP) network.
3 . The method of claim 1 , wherein the first space is associated with at least one of a spatial location and a viewing direction.
4 . The method of claim 3 , wherein the spatial location comprises a three dimensional Cartesian location.
5 . The method of claim 3 , wherein the viewing direction comprises a two dimensional viewing direction.
6 . The method of claim 1 , wherein the second space is associated with a volume density and a directional emitted color.
7 . The method of claim 1 , wherein the directional emitted color comprises a Red Green Blue (RGB) color.
8 . A system comprising:
a memory storage; and
a processing unit coupled to the memory storage, wherein the processing unit is operative to:
receive satellite image data;
construct a plurality of nadir rays, each of the nadir rays extending from a unique point of origin to a pixel from the image data, whereby the plurality of nadir rays are parallel to one another, each of the plurality of nadir rays further including a plurality of points, in a first space;
use a Machine Learning (ML) model to map the plurality of points along the plurality of nadir rays in the first space to a second space; and
render the plurality of points along the plurality of nadir rays in the second space into images.
9 . The system of claim 8 , wherein the ML model comprises a Multilayer Perceptron (MLP) network.
10 . The system of claim 8 , wherein the first space is associated with at least one of a spatial location and a viewing direction.
11 . The system of claim 10 , wherein the spatial location comprises a three dimensional Cartesian location.
12 . The system of claim 10 , wherein the viewing direction comprises a two dimensional viewing direction.
13 . The system of claim 8 , wherein the second space is associated with a volume density and a directional emitted color.
14 . The system of claim 8 , wherein the directional emitted color comprises a Red Green Blue (RGB) color.
15 . A non-transitory computer-readable medium that stores a set of instructions which when executed perform a method executed by the set of instructions comprising:
receiving satellite image data;
constructing a plurality of nadir rays, each of the nadir rays extending from a unique point of origin to a pixel from the image data, whereby the plurality of nadir rays are parallel to one another, each of the plurality of nadir rays further including a plurality of points, in a first space;
using a Machine Learning (ML) model to map the plurality of points along the plurality of nadir rays in the first space to a second space; and
rendering the plurality of points along the plurality of nadir rays in the second space into images.
16 . The non-transitory computer-readable medium of claim 15 , wherein the ML model comprises a Multilayer Perceptron (MLP) network.
17 . The non-transitory computer-readable medium of claim 15 , wherein the first space is associated with at least one of a spatial location and a viewing direction.
18 . The non-transitory computer-readable medium of claim 17 , wherein the spatial location comprises a three dimensional Cartesian location.
19 . The non-transitory computer-readable medium of claim 17 , wherein the viewing direction comprises a two dimensional viewing direction.
20 . The non-transitory computer-readable medium of claim 15 , wherein the second space is associated with a volume density and a directional emitted color.