Systems and methods for generation of searchable structures respective of multimedia data content

• US 10,360,253 B2
• Filed: 01/26/2017
• Issued: 07/23/2019
• Est. Priority Date: 10/26/2005
• Status: Active Grant

• Current Assignee: Cortica Ltd.
• Original Assignee: Cortica Ltd.
• Inventors: Raichelgauz, Igal, Odinaev, Karina, Zeevi, Yehoshua Y
• Primary Examiner(s): Vo, Truong V
• Application Number:
  US15/416,415
• Publication Number:
  US 20170139940A1
• Time in Patent Office: 908 Days
• CPC Class Codes: G06F 16/41 Indexing; Data structures therefor; Storage structures

Abstract

A system and method for generating concept structures based on a plurality of multimedia data elements (MMDEs). The method includes: generating, based on the plurality of MMDEs, a plurality of patches, wherein each patch is at least a portion of one of the MMDEs; generating, based on the plurality of patches, a plurality of signatures for the plurality of MMDEs; clustering the generated plurality of signatures into a plurality of clusters; generating metadata for each of the plurality of clusters; and creating, based on the plurality of clusters, at least one concept structure, wherein each concept structure includes at least one of the plurality of clusters and the metadata associated with the at least one of the plurality of clusters.

Claims

1. A method for generating concept structures based on a plurality of multimedia data elements (MMDEs), comprising:

generating, based on the plurality of MMDEs, a plurality of patches, wherein each patch is at least a portion of one of the MMDEs;

generating, based on the plurality of patches, a plurality of signatures for the plurality of MMDEs;

clustering the generated plurality of signatures into a plurality of clusters;

generating metadata for each of the plurality of clusters; and

creating, based on the plurality of clusters, at least one concept structure, wherein each concept structure includes at least one of the plurality of clusters and the metadata associated with the at least one of the plurality of clusters.

2. The method of claim 1, wherein generating the plurality of patches further comprises:

creating, for each MMDE, at least one patch; and

determining, for each created patch, an entropy of the patch, the entropy of each patch representing an amount of interesting information in the patch, wherein the generated plurality of patches includes each created patch having an entropy above a predetermined threshold.

3. The method of claim 2, wherein the entropy of each patch is determined based on at least one of:

at least one sharp edge of the patch, at least one corner of the patch, and at least one border of the patch.

4. The method of claim 2, wherein the entropy of each created patch is determined using a plurality of statistically independent computational cores, wherein the properties of each computational core are set independently of the properties of others of the computational cores.

5. The method of claim 2, wherein each patch has a random size and a random position within the respective MMDE.

6. The method of claim 1, wherein each of the plurality of clusters is a signature reduced cluster (SRC), wherein clustering the generated plurality of signatures further comprises:

determining whether a number of generated patches is above a predetermined threshold; and

creating a plurality of SRCs, when it is determined that the number of generated patches is above the predetermined threshold.

7. The method of claim 6, wherein creating the plurality of signature reduced clusters further comprises:

reducing a number of signatures in each SRC so that each SRC includes only signatures that uniquely identify the SRC, wherein reducing the number of signatures in each SRC further comprises, for each signature in the SRC;

removing the signature from the SRC;

determining if MMDEs associated with the SRC can be recognized using the removed signature SRC; and

adding the removed signature to the SRC, when it is determined that MMDEs associated with the SRC cannot be recognized using the removed signature SRC.

8. The method of claim 6, wherein creating the plurality of SRCs further comprises:

comparing each signature to a plurality of existing clusters to determine whether the signature matches at least one of the existing clusters; and

associating each matching signature with the respective matching at least one of the existing clusters.

9. The method of claim 1, further comprising:

for each cluster, comparing the metadata generated for the cluster to metadata of a plurality of existing clusters to determine whether the metadata of the cluster matches the metadata of at least one of the existing clusters above a predetermined threshold; and

for each cluster having metadata matching at least one of the existing clusters above a predetermined threshold, associating the cluster with the at least one of the existing clusters to form a concept structure.

10. A non-transitory computer readable medium having stored thereon instructions for causing one or more processing units to perform a process, the process comprising:

generating, based on the plurality of MMDEs, a plurality of patches, wherein each patch is at least a portion of one of the MMDEs;

generating, based on the plurality of patches, a plurality of signatures for the plurality of MMDEs;

clustering the generated plurality of signatures into a plurality of clusters; and

creating, based on the plurality of clusters, at least one concept structure, wherein each concept structure includes at least one of the plurality of clusters and the metadata associated with the at least one of the plurality of clusters.

11. A system for generating concept structures from signature reduced clusters (SRCs), comprising:

a processing circuitry; and

a memory, the memory containing instructions that, when executed by the processing circuitry, configure the system to;

generate, based on the plurality of MMDEs, a plurality of patches, wherein each patch is at least a portion of one of the MMDEs;

generate, based on the plurality of patches, a plurality of signatures for the plurality of MMDEs;

cluster the generated plurality of signatures into a plurality of clusters;

generate metadata for each of the plurality of clusters; and

create, based on the plurality of clusters, at least one concept structure, wherein each concept structure includes at least one of the plurality of clusters and the metadata associated with the at least one of the plurality of clusters.

12. The system of claim 11, wherein the system is further configured to:

create, for each MMDE, at least one patch; and

determine, for each created patch, an entropy of the patch, the entropy of each patch representing an amount of interesting information in the patch, wherein the generated plurality of patches includes each created patch having an entropy above a predetermined threshold.

13. The system of claim 12, wherein the entropy of each patch is determined based on at least one of:

at least one sharp edge of the patch, at least one corner of the patch, and at least one border of the patch.

14. The system of claim 12, wherein the entropy of each created patch is determined using a plurality of statistically independent computational cores, wherein the properties of each computational core are set independently of the properties of others of the computational cores.

15. The system of claim 12, wherein each patch has a random size and a random position within the respective MMDE.

16. The system of claim 11, wherein each of the plurality of clusters is a signature reduced cluster (SRC), wherein the system is further configured to:

determine whether a number of generated patches is above a predetermined threshold; and

create a plurality of SRCs, when it is determined that the number of generated patches is above the predetermined threshold.

17. The system of claim 16, wherein the system is further configured to:

reduce a number of signatures in each SRC so that each SRC includes only signatures that uniquely identify the SRC, wherein reducing the number of signatures in each SRC further includes, for each signature in the SRC;

removing the signature from the SRC, determining if MMDEs associated with the SRC can be recognized using the removed signature SRC, and adding the removed signature to the SRC, when it is determined that MMDEs associated with the SRC cannot be recognized using the removed signature SRC.

18. The system of claim 16, wherein the system is further configured to:

compare each signature to a plurality of existing clusters to determine whether the signature matches at least one of the existing clusters; and

associate each matching signature with the respective matching at least one of the existing clusters.

19. The system of claim 11, wherein the system is further configured to:

for each cluster, compare the metadata generated for the cluster to metadata of a plurality of existing clusters to determine whether the metadata of the cluster matches the metadata of at least one of the existing clusters above a predetermined threshold; and

for each cluster having metadata matching at least one of the existing clusters above a predetermined threshold, associate the cluster with the at least one of the existing clusters to form a concept structure.