Claims
- 1. A method of classifying an element in an image into one of a plurality of classification classes, wherein the element has a plurality of features, the method comprising the steps of:
extracting the plurality of features from the image of the element; determining a classification class of the element by at least one of:
selecting and processing a first subgroup of the extracted features to determine a physical characteristic of the element, and selecting and processing a second subgroup of the extracted features in response to the physical characteristic determined to determine a classification class of the element; and selecting and processing a third subgroup of the extracted features to determine a group of classification classes of the element, and selecting and processing a fourth subgroup of the extracted features in response to the determined classification class group to determine a classification class of the element; and modifying the determined classification class of the element based upon a plurality of previously determined classification class determinations.
- 2. The method of claim 1, wherein the element is a biological particle.
- 3. The method of claim 1, wherein each of the determinations includes assigning a probability factor, and further including the step of modifying the determined classification class to an artifact classification in the event one or more of the probability factors used to classify the element fails to exceed a predetermined threshold value.
- 4. The method of claim 1, further comprising the step of:
classifying the element as an artifact based on a physical characteristic of the element, wherein the artifact element bypasses the determination of the classification class of the element.
- 5. The method of claim 1, further comprising the steps of:
determining whether a boundary of the element intersects a border of an image containing the element, and modifying the determined classification class of the element to an artifact classification in the event the element boundary and image border are determined to intersect.
- 6. The method of claim 1, wherein the processing of the first, second, third and fourth subgroups of the extracted features is performed using neural nets.
- 7. The method of claim 6, further comprising the steps of:
training the neural nets by selecting and processing the first, second, third and fourth subgroups of the extracted features using a training set of known elements along with a test set of elements, wherein the training of the neural nets is repeatedly performed until the accuracy rate of the determination of classification class of the test set of elements reaches a predetermined value.
- 8. The method of claim 6, wherein the first, second, third and fourth subgroups of the plurality of features are selected by modifying each of the feature values by a predetermined amount, and selecting those features that affect the output of the respectively neural net the most.
- 9. The method of claim 1, wherein one of the plurality of extracted features is symmetry of the element, and the extraction of the symmetry feature includes:
defining a first line segment that crosses a centroid of the element; defining a second and third line segments for points along the first line segment that orthogonally extend away from the first line segment in opposite directions; utilizing the difference between the lengths of the second and third line segments to calculate the extracted symmetry feature of the element.
- 10. The method of claim 1, wherein one of the plurality of extracted features is skeletonization of the element image, and the extraction of the skeletonization feature includes orthogonally collapsing a boundary of the element to form one or more line segments.
- 11. The method of claim 1, wherein at least one of the plurality of extracted features is a measure of a spatial distribution of the element image, and at least another one of the plurality of extracted features is a measure of a spatial frequency domain of the element image.
- 12. An imaging apparatus for classifying an element in an image into one of a plurality of classification classes, wherein the element has a plurality of features, the apparatus comprising:
means for extracting the plurality of features from the image of the element; means for determining a classification class of the element, the determining means including at least one of:
means for selecting and processing a first subgroup of the extracted features to determine a physical characteristic of the element, and means for selecting and processing a second subgroup of the extracted features in response to the physical characteristic determined to determine a classification class of the element; and means for selecting and processing a third subgroup of the extracted features to determine a group of classification classes of the element, and means for selecting and processing a fourth subgroup of the extracted features in response to the determined classification class group to determine a classification class of the element; and means for modifying the determined classification class of the element based upon a plurality of previously determined classification class determinations.
- 13. The apparatus of claim 12, wherein the element is a biological particle.
- 14. The apparatus of claim 12, wherein each of the determinations includes assigning a probability factor, and wherein the determining means further includes means for modifying the determined classification class to an artifact classification in the event one or more of the probability factors used to classify the element fails to exceed a predetermined threshold value.
- 15. The apparatus of claim 12, further comprising:
means for classifying the element as an artifact based on a physical characteristic of the element, wherein the artifact element bypasses the determining means.
- 16. The apparatus of claim 12, further comprising:
means for determining whether a boundary of the element intersects a border of an image containing the element, and means for modifying the determined classification class of the element to an artifact classification in the event the element boundary and image border are determined to intersect.
- 17. The apparatus of claim 12, wherein the processing of the first, second, third and fourth subgroups of the extracted features is performed using neural nets.
- 18. The apparatus of claim 17, further comprising:
means for training the neural nets by selecting and processing the first, second, third and fourth subgroups of the extracted features using a training set of known elements along with a test set of elements, wherein the training means repeatedly trains the neural nets until the accuracy rate of the determination of classification class of the test set of elements reaches a predetermined value.
- 19. The apparatus of claim 17, wherein the first, second, third and fourth subgroups of the plurality of features are selected by modifying each of the feature values by a predetermined amount, and selecting those features that affect the output of the respectively neural net the most.
- 20. The apparatus of claim 12, wherein one of the plurality of extracted features is symmetry of the element, and the extraction means includes:
means for defining a first line segment that crosses a centroid of the element; means for defining a second and third line segments for points along the first line segment that orthogonally extend away from the first line segment in opposite directions; means for utilizing the difference between the lengths of the second and third line segments to calculate the extracted symmetry feature of the element.
- 21. The apparatus of claim 12, wherein one of the plurality of extracted features is skeletonization of the element image, and the extraction means further includes means for orthogonally collapsing a boundary of the element to form one or more line segments.
- 22. The apparatus of claim 12, wherein at least one of the plurality of extracted features is a measure of a spatial distribution of the element image, and at least another one of the plurality of extracted features is a measure of a spatial frequency domain of the element image.
- 23. A method of classifying an element in an image into one of a plurality of classifications, wherein the element has a plurality of features, the method comprising the steps of:
extracting the plurality of features from the image; determining a classification of the element based upon the plurality of features extracted by a first determination criteria, wherein the first determination criteria includes the classification of the element as an unknown classification; determining a classification of the element by a second determination criteria, different from the first determination criteria, in the event the element is classified as an unknown classification by the first determination criteria; and determining the classification of the element by a third determination criteria, different from the first and second determination criteria, in the event the element is classified as one of a plurality of classifications by the first determination criteria.
- 24. An imaging apparatus for classifying an element in an image into one of a plurality of classification classes, wherein the element has a plurality of features, the apparatus comprising:
an extractor for extracting the plurality of features from the image of the element; a first processor that determines a classification class of the element by at least one of:
selecting and processing a first subgroup of the extracted features to determine a physical characteristic of the element, and selecting and processing a second subgroup of the extracted features in response to the physical characteristic determined to determine a classification class of the element; and selecting and processing a third subgroup of the extracted features to determine a group of classification classes of the element, and selecting and processing a fourth subgroup of the extracted features in response to the determined classification class group to determine a classification class of the element; and a second processor that modifies the determined classification class of the element based upon a plurality of previously determined classification class determinations.
Parent Case Info
[0001] This application claims the benefit of U.S. Provisional Application No. 60/199,237, filed Apr. 24, 2000, and entitled MULTI-NEURAL NET IMAGING APPARATUS AND METHOD.
Provisional Applications (1)
|
Number |
Date |
Country |
|
60199237 |
Apr 2000 |
US |