Detection of change in posture in video
DCFirst Claim
1. A computer readable medium storing computer readable program code for causing a computer to perform a method, comprising the steps of:
- receiving input video data;
detecting a target in the video data;
tracking the target;
identifying a target as a human; and
detecting a change in a posture of the human target in the video data when a change in at least one of a height or aspect ratio of the human target exceeds a predetermined threshold by determining the height or the aspect ratio of the target in a first frame of the video data;
determining the height or aspect ratio of the target in a second frame of the video data;
calculating a difference between the height or aspect ratio of the target in the first and second frames;
detecting the change in the posture when either difference exceeds a predetermined threshold;
determining an elapsed time between the first and second frames; and
detecting the change in the posture when the elapsed time is below a predetermined threshold.
8 Assignments
Litigations
0 Petitions
Reexamination
Accused Products
Abstract
Input video data is processed to detect a change in a posture of a person shown in the video data. The change of posture may be the result of an event, for example, the person falling or getting up. The input video data may include a plurality of frames. Objects in the frames are tracked and then classified, for example, as human and non-human targets. At least one of the position or location of a human target in the frames is identified. Changes in the location or position of the human target between the frames is determined. When the change in at least of the position or location exceeds a predetermined threshold, a falling down event or a getting up event is detected. The changes in the position or location of the human target can be determined based on a number of different factors.
-
Citations
20 Claims
-
1. A computer readable medium storing computer readable program code for causing a computer to perform a method, comprising the steps of:
-
receiving input video data; detecting a target in the video data; tracking the target; identifying a target as a human; and detecting a change in a posture of the human target in the video data when a change in at least one of a height or aspect ratio of the human target exceeds a predetermined threshold by determining the height or the aspect ratio of the target in a first frame of the video data; determining the height or aspect ratio of the target in a second frame of the video data; calculating a difference between the height or aspect ratio of the target in the first and second frames;
detecting the change in the posture when either difference exceeds a predetermined threshold;determining an elapsed time between the first and second frames; and detecting the change in the posture when the elapsed time is below a predetermined threshold. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10)
-
-
11. A computer readable medium storing computer readable program code for causing a computer to perform a method, comprising the steps of:
-
receiving input video data of a place including a plurality of frames; tracking objects in the frames; classifying objects in the frames as human and non-human targets; identifying at least one of a position or location of a human target in the frames by measuring at least one of a height or aspect ratio of the human target; determining changes in the location and position of the human target between frames; and detecting a falling event or a getting up event when the change in at least one of the position or location exceeds a predetermined threshold by calculating a difference in at least one of the height or aspect ratio between frames and determining the change in position or location based on the difference in at least one of the aspect ratio and height determining a trajectory of the human target; predicting at least one of a height and position of the human target at a point in time utilizing calibration information to predict the height or position of the human target; comparing at least one of the predicted height or position to the actual height or position; and detecting the falling event or the getting up event based on at least a difference between the predicted and actual height or position. - View Dependent Claims (12, 13, 14, 15, 16, 17, 18, 19, 20)
-
Specification