Method and apparatus for identifying and quantifying characteristics of seeds and other small objects
First Claim
1. A method for identifying or quantifying one or more characteristics of interest of unknown objects, comprising the steps of:
- A training of a single neural network model with a first and a second training set of known objects having known values for the one or more characteristics of interest;
B validating the optimal neural network model; and
C analyzing unknown objects having unknown values of the one or more characteristics of interest, comprising the steps of;
I imaging the unknown objects having unknown values of the one or more characteristics of interest against a background to obtain an original digital image, wherein the original digital image comprises pixels representing the unknown objects, the background and any debris;
II processing the original digital image to identify, separate, and retain the pixels representing the unknown objects from the pixels representing the background and the pixels representing any debris, and to eliminate the background and any debris;
III analyzing the pixels representing each of the unknown objects to generate data representative of one or more image parameters for each of the unknown objects;
IV providing the data to a chosen flash code deployed from the candidate neural network model;
V analyzing the data through the flash code; and
VI receiving the output data from the flash code in a predetermined format, wherein the output data represents the unknown values of the one or more characteristics of interest of the unknown objects.
1 Assignment
0 Petitions
Accused Products
Abstract
The invention provides a method for identifying or quantifying characteristics of interest of unknown objects, comprising training a single neural network model with training sets of known objects having known values for the characteristics; validating the optimal neural network model; and analyzing unknown objects having unknown values of the characteristics by imaging them to obtain a digital image comprising pixels representing the unknown objects, background and any debris; processing the image to identify, separate, and retain pixels representing the unknown objects from pixels and to eliminate background and debris; analyzing the pixels representing each of the unknown objects to generate data representative of image parameters; providing the data to the flash code deployed from the candidate neural network model; analyzing the data through the flash code; and receiving output data (the unknown values of the characteristics of interest of the unknown objects) from the flash code in a predetermined format.
94 Citations
96 Claims
-
1. A method for identifying or quantifying one or more characteristics of interest of unknown objects, comprising the steps of:
-
A training of a single neural network model with a first and a second training set of known objects having known values for the one or more characteristics of interest;
B validating the optimal neural network model; and
C analyzing unknown objects having unknown values of the one or more characteristics of interest, comprising the steps of;
I imaging the unknown objects having unknown values of the one or more characteristics of interest against a background to obtain an original digital image, wherein the original digital image comprises pixels representing the unknown objects, the background and any debris;
II processing the original digital image to identify, separate, and retain the pixels representing the unknown objects from the pixels representing the background and the pixels representing any debris, and to eliminate the background and any debris;
III analyzing the pixels representing each of the unknown objects to generate data representative of one or more image parameters for each of the unknown objects;
IV providing the data to a chosen flash code deployed from the candidate neural network model;
V analyzing the data through the flash code; and
VI receiving the output data from the flash code in a predetermined format, wherein the output data represents the unknown values of the one or more characteristics of interest of the unknown objects. - View Dependent Claims (82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94)
-
-
2. A method for training of a single neural network model with a first and second training set of known objects having known values for the one or more characteristics of interest, comprising the steps of:
-
A I selecting known objects having known values for the one or more characteristics of interest;
II arranging the known objects into a spectrum according to increasing degree of expression of the one or more characteristics of interest;
III segregating the known objects into a first and a second training set corresponding to a predetermined state of the one or more characteristics of interest;
III imaging each of the first and second training sets against a background to obtain an original digital image for each of the training sets, wherein each of the original digital images comprises pixels representing the known objects, background and any debris;
IV processing the original digital image to identify, separate, and retain the pixels representing the known objects from the pixels representing the background and the pixels representing any debris, and to eliminate the background and any debris;
V analyzing the pixels representing each of the known objects to generate data representative of one or more image parameters for each of the known objects;
VI providing the data to the neural network software to generate multiple candidate neural network models, wherein the multiple candidate neural network models each can have a flash code for deployment; and
VII choosing an optimal neural network model from the multiple candidate neural network models and retaining the corresponding flash code of the optimal neural network model for identifying or quantifying the one or more characteristics of interest of unknown objects having unknown values of the one or more characteristics of interest; and
B validating the optimal neural network model comprising the steps of;
I selecting more than one sample of the known objects having known values for the one or more characteristics of interest;
II imaging each sample against a background to obtain an original digital image for each sample, wherein the original digital image comprises pixels representing the known objects, background and any debris;
III processing the original digital image to identify, separate, and retain the pixels representing the known objects from the pixels representing the background and the pixels representing any debris, and to eliminate the background and any debris;
IV analyzing the pixels representing each of the known objects to generate data representative of one or more image parameters for each of the known objects;
V providing the data to a chosen flash code deployed from the candidate neural network model;
VI analyzing the data through the flash code;
VII evaluating the output data from the flash code for accuracy and repeatability;
VIII choosing and deploying the flash code of the optimal neural network model for identifying or quantifying the one or more characteristics of interest of unknown objects having unknown values of the one or more characteristics of interest.
-
-
3. A method of analyzing unknown objects having unknown values of the one or more characteristics of interest comprising the steps of:
-
I imaging the unknown objects having unknown values of the one or more characteristics of interest against a background to obtain an original digital image, wherein the original digital image comprises pixels representing the unknown objects, the background and any debris;
II processing the original digital image to identify, separate, and retain the pixels representing the unknown objects from the pixels representing the background and the pixels representing any debris, and to eliminate the background and any debris;
III analyzing the pixels representing each of the unknown objects to generate data representative of one or more image parameters for each of the unknown objects;
IV providing the data to the chosen flash code deployed from the candidate neural network model;
V analyzing the data through the flash code; and
VI receiving the output data from the flash code in a predetermined format, wherein the output data represents the unknown values of the one or more characteristics of interest of the unknown objects.
-
- 4. A method of processing a digital image to identify, separate, and retain pixels representing objects from pixels representing the background and pixels representing any debris, and to eliminate the background and any debris.
-
8. A method of processing a digital image comprising pixels representing objects to remove some debris and to separate each of the objects comprising the steps of:
-
i removing some debris from the original digital image of the objects by applying a first digital sieve, wherein the first digital sieve selects the pixels representing each of the objects meeting a predetermined threshold for a first set of one or more image parameters of the objects; and
ii in the image from (i), separating each of the objects that are adjacent by applying an object-splitting algorithm at least once. - View Dependent Claims (9, 10, 11, 12, 13)
-
- 14. A method of processing a digital image comprising pixels representing objects to remove remaining debris or object anomalies comprising the step of separating and removing pixels representing remaining debris or object anomalies from the pixels representing each of the objects by applying a second digital sieve, wherein the second digital sieve selects the pixels representing each of the objects meeting predetermined thresholds for a second set of one or more image parameters.
- 20. A method of analyzing pixels representing objects to generate data representative of one of more parameters for each of the objects wherein the one or more image parameters are dimension, shape, texture, and color.
- 24. A method for obtaining one or more image parameters of color for objects comprising the step of generating an outline of the pixels representing each of the objects.
-
28. A method for identifying or quantifying one or more characteristics of interest of unknown objects comprising the steps of:
-
A training of a single neural network model with a first and a second training set of known objects having known values for the one or more characteristics of interest, wherein training of the single neural network model comprises the steps of;
I selecting known objects having known values for the one or more characteristics of interest;
II arranging the known objects into a spectrum according to increasing degree of expression of the one or more characteristics of interest;
III segregating the known objects into a first and a second training set corresponding to a predetermined state of the one or more characteristics of interest;
IV imaging each of the first and second training sets against a background to obtain an original digital image for each of the training sets, wherein each of the original digital images comprises pixels representing the known objects, background and any debris;
V processing the original digital image to identify, separate, and retain the pixels representing the known objects from the pixels representing the background and the pixels representing any debris, and to eliminate the background and any debris;
VI analyzing the pixels representing each of the known objects to generate data representative of one or more image parameters for each of the known objects;
VII providing the data to the neural network software to generate multiple candidate neural network models, wherein the multiple candidate neural network models each can have a flash code for deployment; and
VIII choosing an optimal neural network model from the multiple candidate neural network models and retaining the corresponding flash code of the optimal neural network model for identifying or quantifying the one or more characteristics of interest of unknown objects having unknown values of the one or more characteristics of interest; and
B validating the optimal neural network model comprising the steps of;
I selecting more than one sample of the known objects having known values for the one or more characteristics of interest;
II imaging each sample against a background to obtain an original digital image for each sample, wherein the original digital image comprises pixels representing the known objects, background and any debris;
III processing the original digital image to identify, separate, and retain the pixels representing the known objects from the pixels representing the background and the pixels representing any debris, and to eliminate the background and any debris;
IV analyzing the pixels representing each of the known objects to generate data representative of one or more image parameters for each of the known objects;
V providing the data to the chosen flash code deployed from the candidate neural network model;
VI analyzing the data through the flash code;
VII evaluating the output data from the flash code for accuracy and repeatability;
VIII choosing and deploying the flash code of the optimal neural network model for identifying or quantifying the one or more characteristics of interest of unknown objects having unknown values of the one or more characteristics of interest; and
C analyzing unknown objects having unknown values of the one or more characteristics of interest, comprising the steps of;
I imaging the unknown objects having unknown values of the one or more characteristics of interest against a background to obtain an original digital image, wherein the original digital image comprises pixels representing the unknown objects, the background and any debris;
II processing the original digital image to identify, separate, and retain the pixels representing the unknown objects from the pixels representing the background and the pixels representing any debris, and to eliminate the background and any debris;
III analyzing the pixels representing each of the unknown objects to generate data representative of one or more image parameters for each of the unknown objects;
IV providing the data to the flash code deployed from the candidate neural network model;
V analyzing the data through the flash code; and
VI receiving the output data from the flash code in a predetermined format, wherein the output data represents the unknown values of the one or more characteristics of interest of the unknown objects. - View Dependent Claims (29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 95, 96)
-
Specification