Systems, devices, and methods for detecting fertility and gender of unhatched eggs

1. A device for detecting a characteristic of an unhatched egg, the device comprising:

a transceiver configured to receive a set of spectral images of an unhatched egg, each of the spectral images obtained in a particular wavelength range, the set of spectral images forming a three-dimensional data cube;

a feature extractor configured to:

combine a first selected subset of the set of spectral images into a first fused image, the first selected subset comprising a different subset than all of the received set of spectral images; and

process the first fused image to:

extract image features from the first fused image, the image features from the first fused image including image texture features that describe spatial and spectral information of the first fused image or a region of the first fused image, the image texture features including an image texture feature having information associated with a spatial distribution of tonal variations within the particular wavelength range; and

combine at least two of the extracted image features from the first fused image into a first feature vector; and

a classifier configured to process the first feature vector to classify the unhatched egg according to at least one characteristic selected from the group of gender of the unhatched egg and fertility of the unhatched non-incubated avian egg;

wherein the transceiver is configured to transmit data signals of results of the classified unhatched egg.

2. The device as claimed in claim 1 , wherein:

the feature extractor is configured to:

combine a second selected subset of the set of spectral images into a second fused image, the second selected subset comprising a different subset than all of the received set of spectral images; and

process the second fused image to:

extract image features from the second fused image, the image features from the second fused image including image texture features that describe spatial and spectral information of the respective first or second fused image or a region of the respective first or second fused image, the image features including an image texture feature having information associated with a spatial distribution of tonal variations within the particular wavelength range;

combine at least two the extracted image features from the second fused image into a second feature vector; and

the classifier is configured to process the first and second feature vectors to classify the unhatched egg according to at least one characteristic, the image features from the first and second feature vectors being classified separately, wherein classification results of the separately classified image features are fused at a decision level by an aggregate classification decision.

3. The device as claimed in claim 1 , wherein the feature extractor is configured to:

at least one of:

select the image features from the first fused image based on at least one of:

a particular characteristic of the at least one characteristic to be detected; or

eggshell color and avian information; and

process the first fused image to detect regions of interest for filtering egg image data from background image data.

4. The device as claimed in claim 1 , wherein said combining the set of images into the first fused image comprises selecting a subset of the images to combine into the first fused image, wherein said selecting the subset of the images is based on at least one of:

a particular characteristic to be detected; or

eggshell color and avian information.

5. The device as claimed in claim 1 , wherein the classifier is configured to provide an indicator of the at least one characteristic, as classified.

6. The device as claimed in claim 1 , wherein:

at least one of:

the image features include image texture features as a set of quantitative measures calculated in image processing to describe the spatial and spectral information of the first fused image or a region of the first fused image; or

the image features are selected from the group consisting of at least one first-order image feature and at least one second-order image feature.

7. The device as claimed in claim 1 , wherein:

at least one of:

the image texture features comprise a second-order measure that reflects a relationship between neighbouring pixels in the fused image or the region of the fused image;

the image texture features comprise a set of quantitative measures calculated in image processing designed to describe spatial and spectral information of the fused image or a region of the fused image; or

the image features are learned by an artificial intelligence system to describe spatial and spectral information of the fused image or a region of the first fused image.

8. The device as claimed in claim 1 , further comprising at least one pre-processor configured to:

at least one of:

determine that the unhatched egg is a white egg or a brown egg for calibration of an imaging system for generating the set of images; or

accept laying avian information and eggshell color for calibration of an imaging system for generating the set of images.

9. The device as claimed in claim 1 , wherein the data signals of the results transmitted by the transceiver are received at an apparatus and used to actuate the apparatus according to at least one characteristic to make physical contact with the unhatched egg to move or separate the classified unhatched egg based on the at least one characteristic.

10. The device as claimed in claim 1 , wherein the set of spectral images of the unhatched egg are generated from an imaging interface for one of:

a spectrograph and a camera; or

a hyperspectral imaging device.

11. A computer-implemented method of detecting a characteristic of an unhatched egg, the method comprising:

receiving, by at least one processor using a transceiver, a set of spectral images of an unhatched egg, each of the spectral images obtained in a particular wavelength range, the set of spectral images forming a three-dimensional data cube;

combining, by the least one processor, a selected subset of the set of spectral images into a first fused image, the selected subset comprising a different subset than all of the received set of spectral images;

processing, by the at least one processor, the first fused image to:

extract image features from the first fused image, the image features from the first fused image including image texture features that describe spatial and spectral information of the first fused image or a region of the first fused image, the image texture features including an image texture feature having information associated with a spatial distribution of tonal variations within the particular wavelength range; and

combine at least two of the extracted image features from the first fused image into a first feature vector;

processing, by the at least one processor, the first feature vector to classify the unhatched egg according to at least one characteristic selected from the group consisting of gender of the unhatched egg and fertility of the unhatched egg; and

transmitting, by the at least one processor using the transceiver, data signals of results of the classified unhatched egg.

12. The method as claimed in claim 11 , comprising:

combining, by the at least one processor, a second selected subset of the set of spectral images into a second fused image, the second selected subset comprising a different subset than all of the received set of spectral images;

processing, by the at least one processor, the second fused image to:

extract image features from the second fused image, the image features from the second fused image including image texture features that describe spatial and spectral information of the second fused image or a region of the second fused image, the image features including an image texture feature having information associated with a spatial distribution of tonal variations within the particular wavelength range;

combine at least two of the extracted image features from the second fused image into a second feature vector and

processing, by the at least one processor, the first and second feature vectors to classify the unhatched egg according to at least one characteristic, the image features from the first and second feature vectors being classified separately, wherein classification results of the separately classified image features are fused at a decision level by an aggregate classification decision.

13. The method as claimed in claim 11 , comprising:

at least one of:

selecting, by the at least one processor, the image features from the first fused image based on at least one of:

a particular characteristic of the at least one characteristic to be detected; or

eggshell color and avian information; and

processing, by the at least one processor, the first fused image to detect regions of interest for filtering egg image data from background image data.

14. The method as claimed in claim 11 , wherein said combining the set of images into the first fused image comprises selecting a subset of the images to combine into the first fused image, wherein said selecting the subset of the images is based on at least one of:

a particular characteristic to be detected; or

eggshell color and avian information.

15. The method as claimed in claim 11 , comprising:

providing, by the at least one processor, an indicator of the at least one characteristic, as classified.

16. The method as claimed in claim 11 , wherein:

at least one of:

the image features include image texture features as a set of quantitative measures calculated in image processing to describe the spatial and spectral information of the first fused image or a region of the first fused image; or

the image features are selected from the group consisting of at least one first-order image feature and at least one second-order image feature.

17. The method as claimed in claim 11 , wherein:

at least one of:

the image texture features comprise a second-order measure that reflects a relationship between neighbouring pixels in the fused image or the region of the fused image;

the image texture features comprise a set of quantitative measures calculated in image processing designed to describe spatial and spectral information of the fused image or a region of the fused image; or

the image features are learned by an artificial intelligence system to describe spatial and spectral information of the fused image or a region of the first fused image.

18. The method as claimed in claim 11 , comprising:

at least one of:

determining, by the at least one processor, that the unhatched egg is a white egg or a brown egg for calibration of an imaging system for generating the set of images; or

accepting, by the at least one processor, laying avian information and eggshell color for calibration of an imaging system for generating the set of images.

19. The method as claimed in claim 11 , comprising:

receiving, at an apparatus, the data signals of the results transmitted by the transceiver; and

using, by the apparatus, the data signals to actuate the apparatus according to at least one characteristic to make physical contact with the unhatched egg to move or separate the classified unhatched egg based on the at least one characteristic.

20. The method as claimed in claim 11 , comprising:

generating the set of spectral images of the unhatched egg from an imaging interface for one of:

a spectrograph and a camera; or

a hyperspectral imaging device.