US5361379A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Soft-decision classifier
Assignee: Boeing Co
•
Filing: 1992-12-09
•
Granted: 1994-11-01
•
Expires: 2012-12-09
Abstract
Abstract An apparatus and method are disclosed for classifying a signal among a plurality of potential signal sources or events in which a signal embodying values for predefined source features is received and conditional likelihoods are determined for the correspondence of each feature value with each of the plurality of events. The individual conditional likelihood values for each feature which corresponds to a common event are multiplied together to form an overall conditional likelihood that any event in question is the source of the signal. The total likelihood products are then sorted into maximum and minimum values, with the maximum likelihood forming a tentative output designation. The ranked values are then examined by a decision logic element to determine if they satisfy certain minimum absolute and relative value threshold limits to verify if the maximum or next higher level likelihood values designate proper output choices. Depending on the results of the threshold tests, one, two, or no signal sources or events are chosen for a signal classification output. At the same time, a conditional probability for the chosen signal is used to indicate a confidence level in the selected choice.
Inventors
- 👤 Stanley A. White
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-26 13:10
CPC Classifications
G; G01; G01V; G01V1/00; G01V1/28; G01V1/288; G01P; G01P1/00; G01P1/07; G01P1/08; G01P1/10; G01P15/00; G01P15/02; G01P15/08; G06; G06F; G06F17/00; G06F17/10; G06F17/18; G08; G08G; G08G1/00; G08G1/01; G08G1/015
Claims Analysis
Total Claims: 17
Claim 1
Independent Claim
291 words
A method for classifying a moving vehicle as falling within one of the least three preselected categories, the method comprising the steps of: (a) sensing, with at least one motion sensor, a plurality of signals from the vehicle; (b) extracting a plurality of feature values from the sensed signals; (c) determining, for each feature value and for each preselected category, a conditional likelihood that the feature value was extracted from a signal from a vehicle in that category; (d) for each category, multiplying together the conditional likelihoods for all feature values to provide a category likelihood; (e) sorting the category likelihoods into a maximum category likelihood, a next maximum category likelihood, and a lower maximum category likelihood; (f) providing, as an output, a designation for the category which corresponds to the maximum category likelihood when that category likelihood exceeds a preselected first threshold and exceeds the next maximum category likelihood by a second preselected threshold; (g) providing, as outputs, multiple designations for categories which correspond to the maximum and next maximum category likelihoods when all three of the following conditions obtain: (1) the maximum category likelihood exceeds the first threshold; (2) the maximum category likelihood does not exceed the next maximum category likelihood by the second threshold; and (3) the next maximum category likelihood does exceed the lower maximum category likelihood by the second threshold; and (h) providing, as an output, no designation for a category when either of the following conditions obtains: (1) the maximum category likelihood does not exceed the first threshold; or (2) the maximum category likelihood does not esceed the next maximum category likelihood by the second preselected threshold, and the next maximum category likelihood does exceed the lower maximum category likelihood by the second preselected threshold.
Claim 2
Independent Claim
291 words
A method for classifying a moving vehicle as falling within one of the least three preselected categories, the method comprising the steps of: (a) sensing, with at least one motion sensor, a plurality of signals from the vehicle; (b) extracting a plurality of feature values from the sensed signals; (c) determining, for each feature value and for each preselected category, a conditional likelihood that the feature value was extracted from a signal from a vehicle in that category; (d) for each category, multiplying together the conditional likelihoods for all feature values to provide a category likelihood; (e) sorting the category likelihoods into a maximum category likelihood, a next maximum category likelihood, and a lower maximum category likelihood; (f) providing, as an output, a designation for the category which corresponds to the maximum category likelihood when that category likelihood exceeds a preselected first threshold and exceeds the next maximum category likelihood by a second preselected threshold; (g) providing, as outputs, multiple designations for categories which correspond to the maximum and next maximum category likelihoods when all three of the following conditions obtain: (1) the maximum category likelihood exceeds the first threshold; (2) the maximum category likelihood does not exceed the next maximum category likelihood by the second threshold; and (3) the next maximum category likelihood does exceed the lower maximum category likelihood by the second threshold; and (h) providing, as an output, no designation for a category when either of the following conditions obtains: (1) the maximum category likelihood does not exceed the first threshold; or (2) the maximum category likelihood does not esceed the next maximum category likelihood by the second preselected threshold, and the next maximum category likelihood does exceed the lower maximum category likelihood by the second preselected threshold.
Claim 3
Dependent Claim
33 words
The method of claim 1, further comprising the step of generating a confidence level for the provided output designation, the confidence level being the sum of the conditional probabilities of the designated categories.
Claim 4
Dependent Claim
73 words
The method of claim 1, wherein the step of determining, for each respective feature, the conditional likelihood that a feature value was extracted from a signal from a vehicle in a predetermined category comprises the step of inputting the feature value as an address to each of a plurality of memory devices, each of which contains conditional likelihoods that the feature value was extracted from a signal from a vehicle in that category.
Claim 5
Dependent Claim
117 words
The method of claim 1, wherein the step of providing a designation for the category which corresponds to the maximum category likelihood comprises the steps of: (a) determining if the maximum category likelihood exceeds the first threshold in a first comparison element which provides two-state "yes" and "no" outputs; (b) determining if the maximum category likelihood exceeds the next maximum category likelihood by the second threshold in a second comparison element which provides two-state "yes" and "no" outputs; (c) applying a first logical AND operation to "yes" outputs for the first and second elements; and (d) selecting, as an output, the category which corresponds to the maximum category likelihood when the AND operation generates a logic ONE.
Claim 6
Dependent Claim
100 words
The method of claim 4, wherein said step of providing two designations for the category comprises the steps of: (a) determining if the next maximum category likelihood also exceeds the lower maximum category likelihood by the second preselected threshold in a third comparison element which provides two-state "yes" and "no" outputs; (b) applying a second logical AND operation to "yes" outputs for the first and third comparison elements and a "no" output for the second element; and (c) selecting categories which correspond to the maximum and next maximum category likelihoods as outputs when the AND operation generates a logic ONE.
Claim 7
Dependent Claim
74 words
The method of claim 5, wherein the step of providing "no" designations for the category comprises the steps of: (a) applying a third logical AND operation to "no" outputs for the second and third comparison elements; (b) applying a logic OR operation to the output of the third logic AND operation and the "no" output of the first comparison element; and (c) selecting a "no" output when the OR operation generates a logic ONE.
Claim 8
Dependent Claim
91 words
The method of claim 6, further comprising the step of generating a confidence for the provided output designation, the confidence generating step comprising the steps of: (a) dividing the maximum category likelihood by the sum of all conditional likelihoods to provide a confidence in the designation of the category corresponding to the maximum category likelihood; and (b) dividing a minimum category likelihood by the sum of all conditional likelihoods and subtracting the result from one to provide a confidence in the designation of the categories corresponding to multiple category likelihood values.
Claim 9
Dependent Claim
303 words
Apparatus for classifying a moving vehicle as falling within one of at least three preselected categories, the apparatus comprising: (a) means, comprising at least one motion sensor, for sensing a plurality of signals from the vehicle; (b) means for extracting a plurality of feature values from the sensed signals; (c) means for determining, for each feature value and for each preselected category, a conditional likelihood that the feature value was extracted from a signal from a vehicle in that category; (d) means for, for each category, multiplying together the conditional likelihoods for all feature values to provide a category likelihood; (e) means for sorting the category likelihoods into a maximum category likelihood, a next maximum category likelihood, and a lower maximum category likelihood; (f) means for providing, as an output, a designation for the category which corresponds to the maximum category likelihood when that category likelihood exceeds a preselected first threshold and exceeds the next maximum category likelihood by a second preselected threshold; (g) means for providing, as outputs, multiple designations for categories which correspond to the maximum and next maximum category likelihoods when all three of the following conditions obtain: (1) the maximum category likelihood exceeds the first threshold; (2) the maximum category likelihood does not exceed the next maximum category likelihood by the second threshold; and (3) the next maximum category likelihood does exceed the lower maximum category likelihood by the second threshold; and (h) means for providing, as an output, no designation for a category when either of the following condition obtains: (1) the maximum category likelihood does not exceed the first threshold; or (2) the maximum category likelihood does not exceed the next maximum category likelihood by the second preselected threshold, and the next maximum category likelihood does exceed the lower maximum category likelihood by the second preselected threshold.
Claim 10
Dependent Claim
303 words
Apparatus for classifying a moving vehicle as falling within one of at least three preselected categories, the apparatus comprising: (a) means, comprising at least one motion sensor, for sensing a plurality of signals from the vehicle; (b) means for extracting a plurality of feature values from the sensed signals; (c) means for determining, for each feature value and for each preselected category, a conditional likelihood that the feature value was extracted from a signal from a vehicle in that category; (d) means for, for each category, multiplying together the conditional likelihoods for all feature values to provide a category likelihood; (e) means for sorting the category likelihoods into a maximum category likelihood, a next maximum category likelihood, and a lower maximum category likelihood; (f) means for providing, as an output, a designation for the category which corresponds to the maximum category likelihood when that category likelihood exceeds a preselected first threshold and exceeds the next maximum category likelihood by a second preselected threshold; (g) means for providing, as outputs, multiple designations for categories which correspond to the maximum and next maximum category likelihoods when all three of the following conditions obtain: (1) the maximum category likelihood exceeds the first threshold; (2) the maximum category likelihood does not exceed the next maximum category likelihood by the second threshold; and (3) the next maximum category likelihood does exceed the lower maximum category likelihood by the second threshold; and (h) means for providing, as an output, no designation for a category when either of the following condition obtains: (1) the maximum category likelihood does not exceed the first threshold; or (2) the maximum category likelihood does not exceed the next maximum category likelihood by the second preselected threshold, and the next maximum category likelihood does exceed the lower maximum category likelihood by the second preselected threshold.
Claim 11
Dependent Claim
32 words
The apparatus of claim 8, further comprising means for generating a confidence level for the provided output designation, the confidence level being the sum of the conditional probabilities of the designated categories.
Claim 12
Dependent Claim
91 words
The apparatus of claim 9, wherein the confidence generating means comprises: (a) means for summing all sorted category likelihoods, thereby producing a sum; (b) means for dividing the maximum category likelihood by the sum so as to provide a confidence in the designation of the category corresponding to the maximum category likelihood; (c) means for dividing a minimum category likelihood by the sum, thereby producing a quotient; and (d) means for subtracting the quotient from one to provide a confidence in the designation of the categories corresponding to multiple category likelihoods.
Claim 13
Dependent Claim
59 words
The apparatus of claim 8, wherein the means for determining conditional likelihoods comprises a plurality of nm memory devices where n is the number of features and m is the number of potential events, each memory device containing conditional likelihoods for a feature within one preselected category, each of said memory elements being coupled to receive a feature value.
Claim 14
Dependent Claim
27 words
The apparatus of claim 8, wherein memory devices having conditional likelihoods associated with a common category have their respective outputs connected to inputs of a single multiplier.
Claim 15
Dependent Claim
137 words
The apparatus of claim 8, wherein the means of claim 9(f) comprises: (a) a first comparison element connected to receive the maximum category likelihood and to generate two-state "yes" and "no" outputs dependent upon if the maximum category likelihood exceeds the first threshold; (b) a second comparison element connected to receive the maximum category likelihood and the next maximum category likelihood and to generate two-state "yes" and "no" outputs dependent upon is the maximum category likelihood exceeds the next maximum category likelihood by the second threshold in (c) means connected to the first and second comparison element "yes" outputs for applying a first logical AND operation to the output states; and (d) means for selecting, as an output, the category which corresponds to the maximum category likelihood when the first logical AND operation generates a logic ONE.
Claim 16
Dependent Claim
127 words
The apparatus of claim 13, wherein the means of claim 9(g) comprises: (a) a third comparison element which provides two-state "yes" and "no" outputs connected to receive the next maximum category likelihood and the lower maximum category likelihood and to detect when the next maximum category likelihood also exceeds the lower maximum category likelihood by the second preselected threshold; (b) means connected to the first and third comparison element "yes" outputs and the second element "no" output for applying a second logical AND operation to the output states; and (c) means connected to the means of claim 14(c) and the means of claim 15(b) for selecting categories which correspond to the maximum and next maximum category likelihoods as outputs when the AND operation generates a logic ONE.
Claim 17
Dependent Claim
73 words
The apparatus of claim 14, wherein the means of claim 9(h) comprises: (a) means for applying a third logical AND operation to "no" outputs for the second and third comparison elements; (b) means for applying a logic OR operation to the output of the third logic AND operation and the "no" output of the first comparison element; and (c) means for selecting a "no" output when the OR operation generates a logic ONE.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1997-10-31 | 1998-05-02 | 1998-11-01 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2001-10-31 | 2002-05-02 | 2002-11-01 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2005-10-31 | 2006-05-02 | 2006-11-01 | $7700.00 | Window Closed / Lapsed |