Gateway device for volumetric sensing

1. A system comprising:

a sensor including a set of cameras configured to capture a set of images of an enclosed space;

a processor; and

a non-transitory computer-readable medium containing instructions which, when executed on the processor, cause the processor to perform operations including:

receiving the set of images captured by the set of cameras;

pre-processing the set of images to increase quality and reduce effect of ambient lighting;

building a disparity map for the set of images based at least in part on a relationship between the set of cameras; and

determining an occupancy of the enclosed space based at least in part on the disparity map.

2. The system of claim 1 , wherein pre-processing the set of images includes:

frame aligning the set of images.

3. The system of claim 1 , wherein pre-processing the set of images includes:

performing noise reduction on the set of images to obtain a representative image;

compressing a dynamic range of the captured scene represented in the set of images; and

enhancing image contrast of the representative image.

4. The system of claim 3 , wherein noise reduction and frame alignment are performed concurrently, the noise reduction including determining a running average of pixels in the set of images and the frame alignment including aligning pixels of a first image in the set of images prior to including the first image in the running average.

5. The system of claim 3 , wherein compressing the dynamic range of the captured scene includes:

capturing an overexposed image and an underexposed image and obtaining a fused image using Merten image fusion on the overexposed image and the underexposed image; and

performing dynamic gamma correction on the fused image.

6. The system of claim 1 , wherein pre-processing the set of images includes determining that the set of images have an acceptable noise level, wherein the determination comprises: calculating an average standard deviation of pixel intensity at specific location on the set of images and determining that the standard deviation is below a noise threshold.

7. The system of claim 1 , wherein the set of images includes burst images sampled at a frequency of two images per second.

8. The system of claim 1 , wherein building the disparity map comprises:

calibrating the set of cameras to obtain pitch, yaw, and roll angles associated with the set of cameras;

aligning enhanced images associated with the set of cameras based on the pitch, yaw, and roll angles;

filtering static environment regions from each of the enhanced images; and

performing semi-global block matching using a dynamic image pyramid to obtain the disparity map.

9. The system of claim 1 , wherein executing the instructions on the processor further causes the processor to perform the operations including:

localizing the set of cameras; and

classifying the occupancy of the enclosed space based at least in part on one or more classification algorithms.

10. The system of claim 9 , wherein the one or more classification algorithms includes saturation, stereo degradation, scene motion, scene change, corner matching, or any combination thereof.

11. The system of claim 10 , wherein saturation includes determining a percentage of pixels in a representative image of the set of images that are saturated, and based on the percentage being above a saturation threshold, classifying the occupancy of the enclosed space as full.

12. The system of claim 11 , wherein the saturation threshold is 20 percent.

13. The system of claim 10 , wherein scene motion includes:

blurring a current scene captured image and a previous scene captured image;

determining, from the blurred current scene captured image and the blurred previous scene captured image, whether a scene depicting the enclosed space has changed; and

based on the scene changing, updating the disparity map.

14. The system of claim 1 , wherein the sensor further includes a time of flight sensor, a light sensor, an accelerometer, or any combination thereof.

15. The system of claim 1 , wherein the set of cameras consists of two cameras.

16. A method for determining cargo occupancy comprising:

receiving a set of images of an enclosed space captured by a set of cameras;

pre-processing the set of images to increase quality and reduce effect of ambient lighting;

building a disparity map for the set of images based at least in part on a relationship between the set of cameras; and

determining an occupancy of the enclosed space based at least in part on the disparity map.

17. The method of claim 16 , wherein pre-processing the set of images includes:

performing noise reduction on the set of images to obtain a representative image;

compressing a dynamic range of the captured scene represented in the set of images; and

enhancing image contrast of the representative image.

18. The method of claim 17 , wherein noise reduction and frame alignment are performed concurrently, the noise reduction including determining a running average of pixels in the set of images and the frame alignment including aligning pixels of a first image in the set of images prior to including the first image in the running average.

19. The method of claim 16 , further comprising:

localizing the set of cameras; and

classifying the occupancy of the enclosed space based at least in part on one or more classification algorithms.

20. The method of claim 19 , wherein the one or more classification algorithms includes saturation, stereo degradation, scene motion, scene change, corner matching, or any combination thereof.