Variance-based event clustering for automatically classifying images
First Claim
Patent Images
1. An image classification method in a digital image processing system for automatically classifying a plurality of captured digital images, the method comprising using a digital image processor to perform the steps of:
- receiving a plurality of digital images to be classified, each digital image having associated geographic image capture metadata;
determining a first grouping value according to the associated geographic image capture metadata for each of the plurality of digital images to be classified;
defining the first grouping value for each digital image as a distance between the geographic image capture metadata for each digital image and a relative geographic reference point;
calculating an average of said determined first grouping values;
computing a variance metric of said first grouping values, relative to said average;
determining from said variance metric a first grouping threshold applicable to said first grouping values;
identifying first grouping values beyond said first grouping threshold as group boundaries;
assigning said digital images to a plurality of digital image groups based upon said group boundaries;
determining a second grouping value for each of the plurality of digital images to be classified;
calculating as to one or more of said digital image groups, a group average of said second-grouping values of respective said digital images;
computing a variance metric of respective said second-grouping values relative to each said average;
determining from each said variance metric a respective second-grouping threshold applicable to the respective said digital image group;
identifying ones of said second-grouping values beyond respective said second-grouping thresholds as subgroup boundaries of respective said digital image groups; and
assigning said digital images of each of said one or more digital image groups to a plurality of subgroups based upon respective said subgroup boundaries.
5 Assignments
0 Petitions
Accused Products
Abstract
In an image classification method, a plurality of grouping values are received. The grouping values each have an associated image. An average of the grouping values is calculated. A variance metric of the grouping values, relative to the average is computed. A grouping threshold is determined from the variance metric. Grouping values beyond the grouping threshold are identified as group boundaries. The images are assigned to a plurality of groups based upon the group boundaries.
-
Citations
9 Claims
-
1. An image classification method in a digital image processing system for automatically classifying a plurality of captured digital images, the method comprising using a digital image processor to perform the steps of:
-
receiving a plurality of digital images to be classified, each digital image having associated geographic image capture metadata; determining a first grouping value according to the associated geographic image capture metadata for each of the plurality of digital images to be classified; defining the first grouping value for each digital image as a distance between the geographic image capture metadata for each digital image and a relative geographic reference point; calculating an average of said determined first grouping values; computing a variance metric of said first grouping values, relative to said average; determining from said variance metric a first grouping threshold applicable to said first grouping values; identifying first grouping values beyond said first grouping threshold as group boundaries; assigning said digital images to a plurality of digital image groups based upon said group boundaries; determining a second grouping value for each of the plurality of digital images to be classified; calculating as to one or more of said digital image groups, a group average of said second-grouping values of respective said digital images; computing a variance metric of respective said second-grouping values relative to each said average; determining from each said variance metric a respective second-grouping threshold applicable to the respective said digital image group; identifying ones of said second-grouping values beyond respective said second-grouping thresholds as subgroup boundaries of respective said digital image groups; and assigning said digital images of each of said one or more digital image groups to a plurality of subgroups based upon respective said subgroup boundaries. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8)
-
-
9. An image classification method in a digital image processing system for automatically classifying a plurality of captured digital images, the method comprising capturing a plurality of digital images and using a digital image processor to perform the steps of:
-
receiving a grouping value for each of the plurality of captured digital images; calculating an average of said grouping values; computing a variance metric of said grouping values, relative to said average; determining from said variance metric a grouping threshold applicable to said grouping values; identifying grouping values beyond said grouping threshold as group boundaries; assigning said digital images to a plurality of digital image groups based upon said group boundaries; and wherein said grouping threshold is expressed by the equation;
event threshold=0.2+8.159e(−
0.0002*(s^2))where e is the natural logarithm, and s is the standard deviation of said grouping values.
-
Specification