System and method for hyperspectral image processing to identify foreign object

1. A system comprising:

a memory; and

at least one processor to:

generate a synthetic training set comprising a plurality of hyperspectral images, the plurality of hyperspectral images based on actual hyperspectral images of a food object, synthetic hyperspectral images of the food object that include data added to the actual hyperspectral images of the food object, and synthetic hyperspectral images of the food object rendered in one of a different position and orientation;

acquire a hyperspectral image of the food object by an imaging device, the hyperspectral image of the food object comprising a three-dimensional set of images of the food object, each image in the set of images representing the food object in a wavelength range of the electromagnetic spectrum;

normalize the hyperspectral image of the food object;

select a region of interest in the hyperspectral image, the region of interest comprising a subset of at least one image in the set of images;

extract spectral features from the region of interest in the hyperspectral image; and

compare the spectral features from the region of interest with the plurality of hyperspectral images in the synthetic training set to determine particular characteristics of the food object and determine that the hyperspectral image indicates a foreign object.

2. The system of claim 1 , the at least one processor further to train an object model using data comprising the plurality of hyperspectral images in the synthetic training set and ground truth measurements associated with the food object and determine the particular characteristics of the food object based on the object model.

3. The system of claim 1 , the at least one processor further to determine that the foreign object comprises at least one of plastic, paper, a metal hook, a cigarette butt, cardboard, and textile.

4. The system of claim 1 , wherein the imaging device comprises a hyperspectral imaging device and the system further comprises at least one illumination device that illuminates the food object and a conveyor belt that moves the food object past the imaging device as the hyperspectral imaging device captures the hyperspectral image.

5. The system of claim 1 , wherein the food object comprises one of an avocado, a banana, an orange, a peach, an onion, a pear, a kiwi fruit, a mango, berries, a tomato, an apple, sugar, pork, lamb, beef, and fish.

6. The system of claim of claim 1 , the at least one processor further to determine the foreign object located on the food object.

7. The system of claim 1 , the at least one processor further to augment at least one image of the plurality of hyperspectral images in the synthetic training set by inserting the foreign object.

8. A method comprising:

generating, by a processor, a synthetic training set comprising a plurality of hyperspectral images, the plurality of hyperspectral images based on actual hyperspectral images of a food object, synthetic hyperspectral images of the food object that include data added to the actual hyperspectral images of the food object, and synthetic hyperspectral images of the food object rendered in one of a different position and orientation;

acquiring, by the processor, a hyperspectral image of the food object by an imaging device, the hyperspectral image of the food object comprising a three-dimensional set of images of the food object, each image in the set of images representing the food object in a wavelength range of the electromagnetic spectrum;

normalizing, by the processor, the hyperspectral image of the food object;

selecting, by the processor, a region of interest in the hyperspectral image, the region of interest comprising a subset of at least one image in the set of images;

extracting, by the processor, spectral features from the region of interest in the hyperspectral image; and

comparing, by the processor, the spectral features from the region of interest with the plurality of hyperspectral images in the synthetic training set to determine particular characteristics of the food object and determine that the hyperspectral image indicates a foreign object.

9. The method of claim 8 , further comprising training an object model using data comprising the plurality of hyperspectral images in the synthetic training set and ground truth measurements associated with the food object and determining the particular characteristics of the food object based on the object model.

10. The method of claim 8 , further comprising determining that the foreign object comprises at least one of plastic, paper, a metal hook, a cigarette butt, cardboard, and textile.

11. The method of claim 8 , the imaging device comprising a hyperspectral imaging device and the method further comprising illuminating the food object by at least one illumination device and moving the object past the imaging device by a conveyor belt as the hyperspectral imaging device captures the hyperspectral image.

12. The method of claim 8 , wherein the food object comprises one of an avocado, a banana, an orange, a peach, an onion, a pear, a kiwi fruit, a mango, berries, a tomato, an apple, sugar, pork, lamb, beef, and fish.

13. The method of claim 8 , further comprising determining the foreign object located on the food object.

14. The method of claim 8 , further comprising augmenting at least one image of the plurality of hyperspectral images in the synthetic training set by inserting the foreign object.

15. A non-transitory computer-readable storage medium, having instructions stored thereon that, when executed by a computing device cause the computing device to perform operations, the operations comprising:

generating a synthetic training set comprising a plurality of hyperspectral images, the plurality of hyperspectral images based on actual hyperspectral images of a food object, synthetic hyperspectral images of the food object that include data added to the actual hyperspectral images of the food object, and synthetic hyperspectral images of the food object rendered in one of a different position and orientation;

acquiring a hyperspectral image of the food object by an imaging device, the hyperspectral image of the food object comprising a three-dimensional set of images of the food object, each image in the set of images representing the food object in a wavelength range of the electromagnetic spectrum;

normalizing the hyperspectral image of the food object;

selecting a region of interest in the hyperspectral image, the region of interest comprising a subset of at least one image in the set of images;

extracting spectral features from the region of interest in the hyperspectral image; and

comparing the spectral features from the region of interest with the plurality of hyperspectral images in the synthetic training set to determine particular characteristics of the food object and determine that the hyperspectral image indicates a foreign object.

16. The non-transitory computer-readable storage medium of claim 15 , the operations further comprising training an object model using data comprising the plurality of hyperspectral images in the synthetic training set and ground truth measurements associated with the food object and determining the particular characteristics of the food object based on the object model.

17. The non-transitory computer-readable storage medium of claim 15 , the operations further comprising determining that the foreign object comprises at least one of plastic, paper, a metal hook, a cigarette butt, cardboard, and textile.

18. The non-transitory computer-readable storage medium of claim 15 , the imaging device comprising a hyperspectral imaging device and the operations further comprising illuminating the food object by at least one illumination device and moving the object past the imaging device by a conveyor belt as the hyperspectral imaging device captures the hyperspectral image.

19. The non-transitory computer-readable storage medium of claim 15 , wherein the food object comprises one of an avocado, a banana, an orange, a peach, an onion, a pear, a kiwi fruit, a mango, berries, a tomato, an apple, sugar, pork, lamb, beef, and fish.

20. The non-transitory computer-readable storage medium of claim 15 , the operations further comprising determining the foreign object located on the food object.