Method and apparatus for multi-dimensional content search and video identification
A multi-dimensional database and indexes and operations on the multi-dimensional database are described which include video search applications or other similar sequence or structure searches. Traversal indexes utilize highly discriminative information about images and video sequences or about object shapes. Global and local signatures around keypoints are used for compact and robust retrieval and discriminative information content of images or video sequences of interest. For other objects or structures relevant signature of pattern or structure are used for traversal indexes. Traversal indexes are stored in leaf nodes along with distance measures and occurrence of similar images in the database. During a sequence query, correlation scores are calculated for single frame, for frame sequence, and video clips, or for other objects or structures.
1. A computer-implemented method for storing information associated with a video sequence in a reference database, the computer-implemented method comprising:
obtaining, by a processor, data associated with the video sequence; and
for each frame of a set of frames of the video sequence:
determining, by the processor, respective global features of a global region of interest,
determining, by the processor, respective local features of a respective keypoint within the global region of interest, wherein, for multiple frames of the set of frames of the video sequence, the respective keypoints correspond to different respective locations within the global region of interest,
generating, by the processor, a respective signature using both the global features for the frame and the local features for the frame,
determining, by the processor, a respective hash value for the frame based on the signature for the frame, and
storing, by the processor, the data associated with the video sequence in the reference database in association with the hash value for the frame.
2. The computer-implemented method of claim 1 , wherein the global region of interest comprises multiple sectors, and wherein the global features comprise features of respective sectors of the multiple sectors.
3. The computer-implemented method of claim 2 , wherein determining the global features comprises comparing at least one feature of each of one or more of the multiple sectors to a threshold value to generate a value for the sector.
4. The computer-implemented method of claim 2 , wherein the global region of interest is a rectangular grid.
5. The computer-implemented method of claim 1 , further comprising selecting the keypoint based on features of the global region of interest.
6. The computer-implemented method of claim 1 , wherein the local features comprise features of an area around the keypoint.
7. The computer-implemented method of claim 6 , wherein determining the local features comprises dividing the area around the keypoint into sectors, wherein the local features comprise features of respective sectors.
8. The computer-implemented method of claim 1 , further comprising selecting the frame based on a comparison of the frame with other frames of the video sequence.
9. The computer-implemented method of claim 1 , wherein the associated data comprises a video name and an indication of a frame number of the frame.
10. A multi-dimensional content search system comprising:
one or more processors; and
a computer-readable medium having stored therein instructions that are executable by the one or more computers to cause the multi-dimensional content search system to perform functions comprising:
obtaining data associated with a video sequence, and
for each frame of a set of frames of the video sequence:
determining respective global features of a global region of interest;
determining respective local features of a respective keypoint within the global region of interest, wherein, for multiple frames of the set of frames of the video sequence, the respective keypoints correspond to different respective locations within the global region of interest;
generating a respective signature using both the global features for the frame and the local features for the frame;
determining a respective hash value for the frame based on the signature for the frame; and
storing the data associated with the video sequence in a reference database in association with the hash value for the frame.
11. The multi-dimensional content search system of claim 10 , wherein the global region of interest comprises multiple sectors, and wherein the global features comprise features of respective sectors of the multiple sectors.
12. The multi-dimensional content search system of claim 11 , wherein determining the global features comprises comparing at least one feature of each of one or more of the multiple sectors to a threshold value to generate a value for the sector.
13. The multi-dimensional content search system of claim 11 , wherein the global region of interest is a rectangular grid.
14. The multi-dimensional content search system of claim 10 , wherein the functions further comprise selecting the keypoint based on features of the global region of interest.
15. The multi-dimensional content search system of claim 10 , wherein the local features comprise features of an area around the keypoint.
16. The multi-dimensional content search system of claim 15 , wherein determining the local features comprises dividing the area around the keypoint into sectors, wherein the local features comprise features of respective sectors of the sectors.
17. A non-transitory computer-readable medium having stored therein instructions that are executable by one or more processors to cause a computing device to perform functions comprising:
obtaining data associated with a video sequence; and
for each frame of a set of frames of the video sequence:
determining respective global features of a global region of interest of the frame,
determining respective local features of a respective keypoint within the global region of interest, wherein, for multiple frames of the set of frames of the video sequence, the respective keypoints correspond to different respective locations within the global region of interest,
generating a respective signature using both the global features for the frame and the local features for the frame,
determining a respective hash value for the frame based on the signature for the frame, and
storing the data associated with the video sequence in a reference database in association with the hash value for the frame.
18. The non-transitory computer-readable medium of claim 17 , wherein the global region of interest comprises multiple sectors, and wherein the global features comprise features of respective sectors of the multiple sectors.
19. The non-transitory computer-readable medium of claim 18 , wherein the global region of interest is a rectangular grid.
20. The non-transitory computer-readable medium of claim 17 , wherein the functions further comprise selecting the keypoint based on features of the global region of interest.