Method of automatic management of images in a collection of images and corresponding device
View Patent ↗The present invention relates to the field of management of image data in data storage. In particular, the present invention relates to a method and device for automatic detection of duplicate images in data storage and corresponding device while taking into account user perception of what he considers to be duplicate images, which method and device are particularly efficient with regard to the personalized, automatic management of large amounts of image data.
1. A method of classifying images in an image collection comprising n images, the method comprising:
classifying transformed images in image sets according to perceived resemblance, wherein the transformed images are based on a same reference image, and wherein each of the transformed images has a distinct fingerprint distance with regard to the reference image;
determining at least one fingerprint distance threshold that delimits the determined image sets within fingerprint distance zones;
classifying the images in the image collection, comprising, for each image i in the image collection;
determining of a fingerprint distance between said image i and each of other n−1 images in the image collection; and
classifying said each of other n−1 images in one of said fingerprint distance zones according to the determined fingerprint distance.
2. The method according to claim 1 , further comprising:
associating an action with each of the fingerprint distance zones; and
for each image classified in one of said fingerprint distance zones, executing an action associated with said fingerprint distance zone.
3. The method according to claim 1 , wherein:
said at least one fingerprint distance threshold comprises at least a first and at least a second fingerprint distance threshold, said first fingerprint distance threshold limiting a first fingerprint distance zone inside which is a set of transformed images classified in an identical and slightly modified image set; and
said second fingerprint distance threshold delimits a second zone outside said first zone and inside of which are transformed images classified in a largely modified image set and outside of which are transformed images classified in an image set of different images.
4. The method according to claim 3 , further comprising:
choosing a fingerprint distance determination method for determining in which of said fingerprint distance zones an image falls, as a function of a position of said first and second thresholds relative to a position of a third threshold of fingerprint distance that determines a fingerprint distance zone inside which a first fingerprint determination method is used, and outside of which a second fingerprint determination method is used.
5. The method according to claim 4 , wherein said first fingerprint determination method is according to a local description approach, and said second fingerprint method is according to a global description approach.
6. The method according to claim 1 , wherein said fingerprint distance is determined according to Euclidian-based distances.
7. An image management device for classifying images in an image collection comprising n images, the device comprising:
at least one processor; and
a memory coupled to the at least one processor,
the at least one processor configured to:
classify transformed images in image sets according to perceived resemblance, wherein the transformed images are based on a same reference image, and wherein each of the transformed images has a distinct fingerprint distance with regard to the reference image;
determine at least one fingerprint distance threshold that delimits the determined image sets within fingerprint distance zones;
classify the images in the image collection, and for each image i in the image collection;
determine a fingerprint distance between said image i and each of other n−1 images in the image collection; and
classify each of other n−1 images in one of said fingerprint distance zones according to the determined fingerprint distance.
8. The device according to claim 7 , wherein the at least one processor is further configured to:
associate an action with each of the fingerprint distance zones; and
execute, for each image classified in one of said fingerprint distance zones, an action associated with said fingerprint distance zone.
9. The device according to claim 7 , wherein:
said at least one fingerprint distance threshold comprises at least a first and at least a second fingerprint distance threshold, said first fingerprint distance threshold limiting a first fingerprint distance zone inside which is a set of transformed images classified in an identical and slightly modified image set; and
said second fingerprint distance threshold delimits a second zone outside said first zone and inside of which are transformed images classified in a largely modified image set and outside of which are transformed images classified in an image set of different images.
10. The device according to claim 9 , wherein the at least one processor is further configured to:
choose a fingerprint distance determination method for determining in which of said fingerprint distance zones an image falls, as a function of a position of said first and second thresholds relative to a position of a third threshold of fingerprint distance that determines a fingerprint distance zone inside which a first fingerprint determination method is used, and outside of which a second fingerprint determination method is used.