Produce recognition system and method including weighted rankings

• US 6,412,694 B1
• Filed: 09/20/2000
• Issued: 07/02/2002
• Est. Priority Date: 09/20/2000
• Status: Expired due to Term

• Current Assignee: NCR Voyix Corporation
• Original Assignee: NCR Corporation (NCR Voyix Corporation)
• Inventors: Kumar, Alok
• Primary Examiner(s): Lee, Michael G.
• Assistant Examiner(s): St. Cyr, Daniel
• Application Number:
  US09/666,737
• Time in Patent Office: 650 Days
• Field of Search: 235/378, 235/385, 235/383, 235/462.01, 235/462.11, 235/462.14, 235/462.41, 235/462.42, 705/20, 705/23, 705/24
• US Class Current: 235/462.01
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06K 7/10861 sensing of data fields affixed to objects or articles, e.g. coded labels postal sorting B07C3/14, conveying articles B65G47/48

Abstract

A produce recognition system and method which uses weighted rankings to increase the possibility that candidate identifications for a produce item will appear in the first-displayed screen of candidate identifications, rather than later screens accessed through operator scrolling. The system in includes a produce data collector which collects produce data for a produce item, and a computer system which compares the produce data to reference produce data to obtain a list of candidate identifications with first rankings based upon probability of match, determining weights for the candidate identifications based upon numbers of times the candidate identifications were correct identifications and were displayed in predetermined operator selection screen locations during past recognition events, applies the weights to the candidate identifications to obtain second rankings, displays the candidate identifications in order of the second rankings, and records an operator selection of one of the candidate identifications.

Claims

1. A method of identifying a produce item comprising the steps of:

collecting produce data from the produce item;

comparing the produce data to reference produce data to obtain a list of candidate identifications with first rankings based upon probability of match;

determining weights for the candidate identifications based upon numbers of times the candidate identifications were correct identifications and were displayed in predetermined operator selection screen locations during past recognition events;

applying the weights to the candidate identifications to obtain second rankings;

displaying the candidate identifications in order of the second rankings; and

recording an operator selection of one of the candidate identifications.

2. The method as recited in claim 1, wherein the produce data comprises spectral data.

3. A method of identifying a produce item comprising the steps of:

capturing spectral data from the produce item;

comparing the spectral data to reference spectral data to produce a list of first candidate identifications with first rankings based upon probability of match;

determining weights for second candidate identifications within the list of first candidate identifications whose corresponding produce items are more difficult to identify through spectral analysis than other produce items;

applying the weights to the second candidate identifications;

determining second rankings for first candidate identifications;

displaying the first candidate identifications in order of the second rankings; and

recording an operator selection of one of the candidate identifications.

4. A method of identifying a produce item comprising the steps of:

capturing spectral data from the produce item;

comparing the produce data to reference spectral data to produce a list of candidate identifications with first rankings based upon probability of match;

increasing the rankings of a number of the candidate identifications whose corresponding produce items are more difficult to identify through spectral analysis than other produce items to produce second rankings;

displaying the candidate identifications in order of the second rankings; and

recording an operator selection of one of the candidate identifications.

5. A method of displaying candidate identifications for a produce item for operator selection comprising the steps of:

determining first rankings for the candidate identifications from a comparison of collected produce data with reference produce data under a produce recognition method;

increasing the rankings of a number of the candidate identifications whose corresponding produce items are more difficult to identify than other produce items using the produce recognition method to produce second rankings;

displaying the candidate identifications in order of the second rankings; and

recording an operator selection of one of the candidate identifications.

6. The method as recited in claim 5, wherein the produce data comprises spectral data.

7. A method of displaying candidate identifications for a produce item for operator selection comprising the steps of:

determining first rankings for the candidate identifications from a comparison of collected produce data with reference produce data;

determining weights for the candidate identifications based upon numbers of times the candidate identifications were correct identifications and were displayed in predetermined operator selection screen locations during previous recognition events;

applying the weights to the candidate identifications to obtain second rankings; and

displaying the candidate identifications in order of the second rankings.

8. A method of displaying candidate identifications for a subject produce item for operator selection comprising the steps of:

determining first rankings for the candidate identifications from a comparison of collected produce data with reference produce data under a produce recognition method;

determining a number of choice locations for displaying the candidate identifications;

obtaining past recognition data including numbers of times correct candidate identifications for produce items associated with the candidate identifications have appeared in each of the choice locations in past recognition events;

determining expected locations within the choice locations for displaying the correct candidate identifications from the number of choice locations and the past recognition data;

determining weights for the candidate identifications from the expected locations and the number of choice locations;

applying the weights to the candidate identifications to obtain second rankings; and

displaying the candidate identifications in order of the second rankings.

9. A produce recognition system comprising:

a produce data collector which collects produce data for a produce item; and

a computer system which compares the produce data to reference produce data to obtain a list of candidate identifications with first rankings based upon probability of match, determining weights for the candidate identifications based upon numbers of times the candidate identifications were correct identifications and were displayed in predetermined operator selection screen locations during past recognition events, applies the weights to the candidate identifications to obtain second rankings, displays the candidate identifications in order of the second rankings, and records an operator selection of one of the candidate identifications.

10. A produce recognition system comprising:

a produce data collector which collects produce data for a produce item; and

a computer system which compares the produce data to reference produce data to obtain a list of candidate identifications with first rankings based upon probability of match under a produce recognition method, increases the first rankings of a number of the candidate identifications whose corresponding produce items are more difficult to identify than other produce items using the produce recognition method to produce second rankings, displays the candidate identifications in order of the second rankings, and records an operator selection of one of the candidate identifications.

11. The produce recognition system as recited in claim 10, wherein the produce data collector comprises a spectrometer.

12. A produce recognition system comprising:

a produce data collector which collects produce data for a subject produce item; and

a computer system which compares the produce data to reference produce data to obtain a list of candidate identifications, determines first rankings for the candidate identifications based upon probability of match, determines a number of choice locations for displaying the candidate identifications, obtains past recognition data including numbers of times correct candidate identifications for produce items associated with the candidate identifications have appeared in each of the choice locations in past recognition events, determines expected locations within the choice locations for displaying the correct candidate identifications from the number of choice locations and the past recognition data, determines weights for the candidate identifications from the expected locations and the number of choice locations, applies the weights to the candidate identifications to obtain second rankings, and displays the candidate identifications in order of the second rankings.