System and method for image selection using multivariate time series analysis
First Claim
1. An image selection method comprising:
- receiving a collection of images;
optionally, filtering the collection of images, the filtering comprising removing images that are near-duplicates of other images in the collection;
extracting a plurality of features from each of the images in the optionally-filtered collection;
with a processor, considering the optionally-filtered collection of images as an ordered sequence of images and segmenting the sequence of images to form a sequence of segments, each segment comprising at least one image, the segmenting of the sequence of images being based on the extracted features and positions of the images in the sequence of images, the segmenting comprising fitting a step function to the image sequence, the fitting comprising minimizing an error, over all the images in the sequence of images, between normalized values of the features for each image and corresponding values of the step function, the step function being constrained to a maximum number of transition points, each transition point corresponding to a change between step function values; and
selecting images from the segments to form a sub-collection of the images.
1 Assignment
0 Petitions
Accused Products
Abstract
An image selection method includes receiving a collection of images and optionally, filtering the collection of images. The filtering may include removing images that are near-duplicates of other images in the collection. A plurality of features is extracted from each of the images in the optionally-filtered collection. The optionally-filtered collection of images is considered as a time-ordered sequence of images and is segmented to form a sequence of segments. Each segment includes at least one of the images. The segmenting of the sequence of images is based on the extracted features and positions of the images in the sequence of images. Images from the segments are selected to form a sub-collection of the images.
-
Citations
21 Claims
-
1. An image selection method comprising:
-
receiving a collection of images; optionally, filtering the collection of images, the filtering comprising removing images that are near-duplicates of other images in the collection; extracting a plurality of features from each of the images in the optionally-filtered collection; with a processor, considering the optionally-filtered collection of images as an ordered sequence of images and segmenting the sequence of images to form a sequence of segments, each segment comprising at least one image, the segmenting of the sequence of images being based on the extracted features and positions of the images in the sequence of images, the segmenting comprising fitting a step function to the image sequence, the fitting comprising minimizing an error, over all the images in the sequence of images, between normalized values of the features for each image and corresponding values of the step function, the step function being constrained to a maximum number of transition points, each transition point corresponding to a change between step function values; and selecting images from the segments to form a sub-collection of the images. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
-
-
16. An image selection system comprising:
-
an optional filtering component for filtering an input collection of images, the filtering comprising removing images that are near-duplicates of other images in the collection; a features extractor which extracts a plurality of features from images in the optionally-filtered collection; a sequence segmentor which considers the optionally-filtered collection of images as a time-ordered sequence of images and which segments the sequence of images to form a sequence of segments, each segment comprising at least one image, the segmenting of the sequence of images being based on the extracted features and positions of the images in the sequence of images, the segmentation comprising identifying transition points in the sequence of images by optimizing a function which, over all images in the sequence, minimizes an error between normalized values of the features and a step function which is constrained to a maximum number of the transition points; optionally, a representative image selector for selecting images from the segments to form a sub-collection of the images; and a processor which implements at least the features extractor and sequence segmentor. - View Dependent Claims (17, 18, 20, 21)
-
-
19. An image selection method comprising:
-
receiving a collection of images; filtering the collection of images, the filtering comprising removing images that are near-duplicates of other images in the collection; extracting features from the images in the filtered collection; and with a processor, considering the images as forming a time-ordered sequence and segmenting the sequence of images to form a sequence of segments, each segment comprising at least one image, the segmenting of the sequence of images being based on the extracted features, the segmentation comprising identifying transition points in the sequence of images by optimizing a function which, over all images in the sequence, minimizes an error between normalized values of the features and a step function which is constrained to a maximum number of the transition points; and providing for selection of images from the segments to form a sub-collection of the images.
-
Specification