IPC분류정보
국가/구분 |
United States(US) Patent
등록
|
국제특허분류(IPC7판) |
|
출원번호 |
US-0347680
(2012-01-10)
|
등록번호 |
US-8660296
(2014-02-25)
|
발명자
/ 주소 |
|
출원인 / 주소 |
|
대리인 / 주소 |
|
인용정보 |
피인용 횟수 :
3 인용 특허 :
6 |
초록
▼
Systems and methods for facilitating video fingerprinting are provided. In one embodiment, a system can include: a memory, a microprocessor, a communication component that receives a video, and a video fingerprinting component that fingerprints the video with a subfingerprint (SFP). The video finger
Systems and methods for facilitating video fingerprinting are provided. In one embodiment, a system can include: a memory, a microprocessor, a communication component that receives a video, and a video fingerprinting component that fingerprints the video with a subfingerprint (SFP). The video fingerprinting component can employ an SFP component stored in the memory and that comprises: a feature extraction component that determines local descriptors for at least one frame of a video; and a quantization component that quantizes the local descriptors to generate first frame information including a set of values for the at least one frame. The SFP component can also include: an accumulation component that accumulates first frame information over a snippet of the video; and an SFP generation component that computes the SFP associated with the snippet. The SFP can be computed based on a hash based on the accumulated first frame information over the snippet.
대표청구항
▼
1. A system comprising: a memory that stores computer executable components; anda microprocessor that executes the following computer executable components stored in the memory: a video fingerprinting component that generates a subfingerprint of a video, the video fingerprinting component comprising
1. A system comprising: a memory that stores computer executable components; anda microprocessor that executes the following computer executable components stored in the memory: a video fingerprinting component that generates a subfingerprint of a video, the video fingerprinting component comprising: a feature extraction component that determines one or more local descriptors for one or more selected frames of the video;a quantization component that quantizes the one or more local descriptors to generate first frame information, the first frame information comprising a set of values for the one or more selected frames of the video;an accumulation component that accumulates the first frame information over a snippet of the video, the snippet comprising selected consecutive ones of the frames of the video; anda subfingerprint generation component that computes the subfingerprint associated with the snippet, wherein the computing the subfingerprint comprises computing a first hash based on an accumulated first frame information over the snippet. 2. The system of claim 1, wherein the subfingerprint generation component is a component that also: generates augmented quantized one or more local descriptors by appending, to quantized one or more local descriptors, an identifier of a part of the snippet in which the quantized one or more local descriptors occurred;computes counts of the augmented quantized one or more local descriptors over the snippet;generates a histogram comprising a weighted set of the augmented quantized one or more local descriptors; andcomputes the subfingerprint based, at least, on computing one or more hash values for the histogram. 3. The system of claim 1, wherein the one or more local descriptors include a local descriptor based on Gabor wavelets. 4. The system of claim 1, wherein the one or more local descriptors include at least one of a scale invariant feature transform (SIFT) descriptor or a Speeded Up Robust Feature (SURF) descriptor. 5. The system of claim 1, wherein the one or more local descriptors include at least one of a type, position, scale, aspect ratio or a descriptor of a portion of the one of the one or more selected frames. 6. The system of claim 1, wherein the one or more selected frames comprises two or more selected frames and the feature extraction component determines the one or more local descriptors for an average of the two or more selected frames. 7. A method, comprising: employing a microprocessor to execute computer executable components stored within a memory to perform the following: determining one or more local descriptors for at least one frame of a video;quantizing the one or more local descriptors to generate first frame information, the first frame information comprising a set of values for the at least one frame;accumulating additional first frame information over a snippet of the video, the snippet comprising the at least one frame and one or more additional frames of the video within a fixed duration of time; andcomputing a subfingerprint associated with the snippet, wherein the computing the subfingerprint comprises computing a first hash based on an accumulated first frame information over the snippet. 8. The method of claim 7, wherein determining one or more local descriptors for at least one frame of the video comprises: determining one or more local descriptors for an average of two or more frames of the video. 9. The method of claim 7, wherein the snippet is separated a predefined amount of time apart from another snippet. 10. The method of claim 7, wherein the one or more local descriptors include a local descriptor based on Gabor wavelets. 11. The method of claim 7, wherein the one or more local descriptors include at least one of a scale invariant feature transform (SIFT) descriptor or a Speeded Up Robust Feature (SURF) descriptor. 12. The method of claim 7, wherein the one or more local descriptors include at least one of a type, position, scale, aspect ratio or a descriptor of a portion of the at least one frame. 13. The method of claim 7, wherein quantizing the one or more local descriptors comprises tree-based quantization of the one or more local descriptors. 14. The method of claim 13, wherein the tree-based quantization comprises a hierarchical k-means quantization. 15. The method of claim 7, wherein quantizing the one or more local descriptors comprises mapping the one or more local descriptors to one or more closest neighbors from a selected set of neighbors. 16. The method of claim 7, wherein computing the subfingerprint comprises: generating augmented quantized one or more local descriptors by appending, to quantized one or more local descriptors, an identifier of a part of the snippet in which the quantized one or more local descriptors occurred;computing counts of the augmented quantized one or more local descriptors over the snippet;generating a histogram comprising a weighted set of the augmented quantized one or more local descriptors; andcomputing the subfingerprint based at least on computing one or more hash values for the histogram. 17. The method of claim 7, wherein the method further comprises: removing selected content from the at least one frame prior to determining the one or more local descriptors. 18. The method of claim 17, wherein the selected content comprises a border. 19. The method of claim 17, wherein the selected content comprises a logo. 20. A method, comprising: employing a microprocessor to execute computer executable components stored within a memory to perform the following: generating a first set of one or more local descriptors for at least one frame of a video;generating a second set of one or more local descriptors for a flipped version of the at least one frame of the video;quantizing the local descriptors of the first set and the second set to generate quantized local descriptors;appending, to each quantized local descriptor, an identifier of a part of a snippet of the video in which the quantized local descriptor is located to generate a set of augmented quantized local descriptors;generating a histogram representing the set of augmented quantized local descriptors; andcomputing a subfingerprint for the video based at least on computing one or more hash values for the histogram. 21. The method of claim 20, wherein generating the first set of one or more local descriptors comprises: generating one or more local descriptors for an average of two or more frames. 22. The method of claim 20, wherein the flipped version is a horizontally flipped version. 23. The method of claim 20, wherein the flipped version is a vertically flipped version. 24. A method for generating a subfingerprint of a snippet of a video, wherein the snippet comprises sets of frames, the method comprising: generating one or more local descriptors for each set of frames;quantizing each local descriptor to yield quantized local descriptors;appending, to each quantized local descriptor, an identifier of a part of the snippet in which the quantized local descriptor is located to yield a set of augmented quantized local descriptors;computing a count of augmented quantized local descriptors to yield a weighted set of augmented quantized local descriptors; andcomputing a hash for the weighted set of augmented quantized local descriptors. 25. The method of claim 24, further comprising: convolving the video with a temporal filter prior to generating the one or more local descriptors. 26. The method of claim 24, further comprising: removing a border from the video prior to generating the one or more local descriptors. 27. The method of claim 24, further comprising: removing a logo from the video prior to generating the one or more local descriptors. 28. The method of claim 24, wherein generating the one or more local descriptors comprises: averaging two or more frames to yield an average of the two or more frames;generating one or more local descriptors for the average. 29. The method of claim 24, wherein quantizing each local descriptor comprises: quantizing a local descriptor using tree-based quantization. 30. The method of claim 24, wherein quantizing each local descriptor comprises: quantizing a position of a local descriptor using two or more grids to generate a quantized position value. 31. The method of claim 30, wherein quantizing each local descriptor further comprises: creating a tuple of quantization values for the local descriptor, the tuple comprising the quantized position value. 32. The method of claim 30, wherein computing the hash comprises computing the hash using a weighted minhash.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.