ANOMALOUS PATTERN DISCOVERY
First Claim
1. A method for anomalous pattern discovery, the method comprising:
- tracking movement of an object in a trajectory in a video data image field, wherein the image field is partitioned into a plurality of different grids;
extracting global image features from the video data image field relative to the tracked object trajectory;
comparing the global image features extracted relative to the tracked object trajectory to a learned trajectory model and generating a global anomaly detection confidence decision value as a function of fitting the extracted global image features to the learned trajectory model;
extracting local image features from the video data for each of the image field grids that include the tracked object trajectory;
comparing the extracted local image features for each of the grids that include the tracked object trajectory to learned feature models for each of the grids that include the tracked object trajectory and generating a local anomaly detection confidence decision value for each of said grids that include the tracked object trajectory as a function of fitting the extracted local image features to the learned feature models for the each grids; and
fusing the generated global anomaly detection confidence decision value with the generated local anomaly detection confidence decision values for the grids that include the tracked object trajectory into a fused anomaly decision with respect to the tracked object.
2 Assignments
0 Petitions
Accused Products
Abstract
A trajectory of movement of an object is tracked in a video data image field that is partitioned into a plurality of different grids. Global image features from video data relative to the trajectory are extracted and compared to a learned trajectory model to generate a global anomaly detection confidence decision value as a function of fitting to the learned trajectory model. Local image features are also extracted for each of the image field grids that include object trajectory, which are compared to learned feature models for the grids to generate local anomaly detection confidence decisions for each grid as a function of fitting to the learned feature models for the grids. The global anomaly detection confidence decision value and the local anomaly detection confidence decision values for the grids are into a fused anomaly decision with respect to the tracked object.
36 Citations
25 Claims
-
1. A method for anomalous pattern discovery, the method comprising:
-
tracking movement of an object in a trajectory in a video data image field, wherein the image field is partitioned into a plurality of different grids; extracting global image features from the video data image field relative to the tracked object trajectory; comparing the global image features extracted relative to the tracked object trajectory to a learned trajectory model and generating a global anomaly detection confidence decision value as a function of fitting the extracted global image features to the learned trajectory model; extracting local image features from the video data for each of the image field grids that include the tracked object trajectory; comparing the extracted local image features for each of the grids that include the tracked object trajectory to learned feature models for each of the grids that include the tracked object trajectory and generating a local anomaly detection confidence decision value for each of said grids that include the tracked object trajectory as a function of fitting the extracted local image features to the learned feature models for the each grids; and fusing the generated global anomaly detection confidence decision value with the generated local anomaly detection confidence decision values for the grids that include the tracked object trajectory into a fused anomaly decision with respect to the tracked object. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10)
-
-
11. A method of providing a service for anomalous pattern discovery, the method comprising:
-
providing an object detector and tracker that tracks movement of an object in a trajectory in a video data image field, wherein the image field is partitioned into a plurality of different grids; providing a global trajectory feature extractor that extracts global image features from the video data image field relative to the object trajectory; providing a global anomaly decider that compares the global image features extracted relative to the object trajectory to a learned trajectory model and generates a global anomaly detection confidence decision value as a function of a fit of the extracted global image features to the learned trajectory model; providing a local grid feature extractor that extracts features from the video data for each of the image field grids that include the object trajectory; providing a local anomaly decider that compares the extracted local image features for each of the grids that include the object trajectory to learned feature models for each of the grids that include the object trajectory and generates a local anomaly detection confidence decision value for each of said grids that include the object trajectory as a function of a fit of the extracted local image features to the learned feature models for the each grids; and a local-global decision fuser that fuses the generated global anomaly detection confidence decision value with the generated local anomaly detection confidence decision values for the grids that include the object trajectory into a fused anomaly decision with respect to the tracked object. - View Dependent Claims (12, 13, 14, 15)
-
-
16. A system, comprising:
-
a processing unit, computer readable memory and a computer readable storage medium; first program instructions to track movement of an object in a trajectory in a video data image field, wherein the image field is partitioned into a plurality of different grids; second program instructions to extract global image features from the video data image field relative to the tracked object trajectory, and to compare the global image features extracted relative to the tracked object trajectory to a learned trajectory model and generate a global anomaly detection confidence decision value as a function of fitting the extracted global image features to the learned trajectory model; third program instructions to extract local image features from the video data for each of the image field grids that include the tracked object trajectory, and to compare the extracted local image features for each of the grids that include the tracked object trajectory to learned feature models for each of the grids that include the tracked object trajectory and generate a local anomaly detection confidence decision value for each of said grids that include the tracked object trajectory as a function of fitting the extracted local image features to the learned feature models for the each grids; and fourth program instructions to fuse the generated global anomaly detection confidence decision value with the generated local anomaly detection confidence decision values for the grids that include the tracked object trajectory into a fused anomaly decision with respect to the tracked object; wherein the first, second, third and fourth program instructions are stored on the computer readable storage medium for execution by the processing unit via the computer readable memory. - View Dependent Claims (17, 18, 19, 20)
-
-
21. An article of manufacture, comprising:
-
a computer readable storage medium having computer readable program code embodied therewith, the computer readable program code comprising instructions that, when executed by a computer processor, cause the computer processor to; track movement of an object in a trajectory in a video data image field, wherein the image field is partitioned into a plurality of different grids; extract global image features from the video data image field relative to the tracked object trajectory; compare the global image features extracted relative to the tracked object trajectory to a learned trajectory model; generate a global anomaly detection confidence decision value as a function of fitting the extracted global image features to the learned trajectory model; extract local image features from the video data for each of the image field grids that include the tracked object trajectory; compare the extracted local image features for each of the grids that include the tracked object trajectory to learned feature models for each of the grids that include the tracked object trajectory; generate a local anomaly detection confidence decision value for each of said grids that include the tracked object trajectory as a function of fitting the extracted local image features to the learned feature models for the each grids; and fuse the generated global anomaly detection confidence decision value with the generated local anomaly detection confidence decision values for the grids that include the tracked object trajectory into a fused anomaly decision with respect to the tracked object. - View Dependent Claims (22, 23, 24, 25)
-
Specification