Methods, devices and systems for detecting objects in a video
First Claim
Patent Images
1. A method of detecting human objects with a video system, comprising:
- obtaining a first video image from a first video camera;
obtaining a second video image from a second video camera;
determining pixels of the first video image are first foreground pixels and a group of the first foreground pixels constitute a first foreground blob set of one or more first foreground blobs;
for each of plural locations within the first video image, comparing a corresponding predetermined shape with the first foreground blob set to obtain a corresponding first probability of a human at the corresponding location, thereby obtaining plural first probabilities associated with the first video image corresponding to the plural locations within the first video image;
determining pixels of the second video image are second foreground pixels and a group of the second foreground pixels constitute a second foreground blob set of one or more second foreground blobs;
for each of plural locations within the second video image, comparing a corresponding predetermined shape with the second foreground blob set to obtain a corresponding second probability of a human at the corresponding location, thereby obtaining plural second probabilities associated with the second video image corresponding to the plural locations within the second video image;
using the plural first probabilities associated with the first video image and the plural second probabilities associated with the second video image, determining X humans are represented by the first foreground blob set and the second foreground blob set, where X is a whole number; and
providing at least one of a report, an alarm, and an event detection using the determination of the representation of X humans,wherein a size of the corresponding predetermined shapes for each of the plural locations within the first video image and each of the plural locations within the second video image is determined in response to calibration of the video system.
4 Assignments
0 Petitions
Accused Products
Abstract
Methods, devices and systems for performing video content analysis to detect humans or other objects of interest a video image is disclosed. The detection of humans may be used to count a number of humans, to determine a location of each human and/or perform crowd analyses of monitored areas.
-
Citations
38 Claims
-
1. A method of detecting human objects with a video system, comprising:
-
obtaining a first video image from a first video camera; obtaining a second video image from a second video camera; determining pixels of the first video image are first foreground pixels and a group of the first foreground pixels constitute a first foreground blob set of one or more first foreground blobs; for each of plural locations within the first video image, comparing a corresponding predetermined shape with the first foreground blob set to obtain a corresponding first probability of a human at the corresponding location, thereby obtaining plural first probabilities associated with the first video image corresponding to the plural locations within the first video image; determining pixels of the second video image are second foreground pixels and a group of the second foreground pixels constitute a second foreground blob set of one or more second foreground blobs; for each of plural locations within the second video image, comparing a corresponding predetermined shape with the second foreground blob set to obtain a corresponding second probability of a human at the corresponding location, thereby obtaining plural second probabilities associated with the second video image corresponding to the plural locations within the second video image; using the plural first probabilities associated with the first video image and the plural second probabilities associated with the second video image, determining X humans are represented by the first foreground blob set and the second foreground blob set, where X is a whole number; and providing at least one of a report, an alarm, and an event detection using the determination of the representation of X humans, wherein a size of the corresponding predetermined shapes for each of the plural locations within the first video image and each of the plural locations within the second video image is determined in response to calibration of the video system. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27)
-
-
28. A method of detecting human objects with a video system, comprising:
-
obtaining a first video image from a first video camera; obtaining a second video image from a second video camera; determining pixels of the first video image are first foreground pixels and a group of the first foreground pixels constitute a first foreground blob set of one or more first foreground blobs; determining pixels of the second video image are second foreground pixels and a group of the second foreground pixels constitute a second foreground blob set of one or more second foreground blobs; for each of plural locations within the first video image, comparing a corresponding predetermined shape with the first foreground blob set and for each of plural locations within the second video image, comparing a corresponding predetermined shape with the second foreground blob set, to determine X humans are represented by the first foreground blob set and the second foreground blob set, where X is whole number, and to determine a location of each of the X humans within the real world; and providing at least one of a report, an alarm, and an event detection using the determination of the representation of X humans, wherein a size of the corresponding predetermined shapes for each of the plural locations within the first video image and each of the plural locations within the second video image is determined in response to calibration of the video system. - View Dependent Claims (29, 30, 31, 32, 33, 34, 35, 36)
-
-
37. A video surveillance system, comprising:
-
a first video source configured to provide a first video image of a real world scene; a second video source configured to provide a second video image of the real world scene; a foreground detection module configured to detect first foreground pixels of the first video image, a group of the first foreground pixels constituting a first foreground blob set of one or more first foreground blobs, second foreground pixels of the second video image, and a group of the second foreground pixels constituting a second foreground blob set of one or more second foreground blobs; a human detection module configured determine X humans are represented by the first foreground blob set and the second foreground blob set by, for each of plural locations within the first video image, comparing a corresponding predetermined shape with the first foreground blob set and for each of plural locations within the second video image, comparing a corresponding predetermined shape with the second foreground blob set; and a response module configured to provide at least one of a report, an alarm, and an event detection using the determined representation of X humans, wherein the human detection module is configured to associate the plural locations within the first video image with corresponding ones of the plural locations within the second video image based upon determining real world locations that correspond to the plural locations within the first video image and the plural locations within the second video image. - View Dependent Claims (38)
-
Specification