Video Motion Detection, Analysis and Threat Detection Device and Method

• US 20120327241A1
• Filed: 06/24/2011
• Published: 12/27/2012
• Est. Priority Date: 06/24/2011
• Status: Active Grant

• Current Assignee: Ademco Inc. (Resideo Technologies, Inc.)
• Original Assignee: Honeywell International Inc.
• Inventors: Howe, Steven J.
• Granted Patent: US 8,681,223 B2
• Time in Patent Office: Days
• US Class Current: 348/155
• CPC Class Codes: G06V 20/52 Surveillance or monitoring of activities, e.g. for recognising suspicious objects recognising microscopic objects G06V20/69

Abstract

A method and apparatus are provided for detecting human motion. The method includes the steps of a camera collecting video images of a secured area, extracting objects in motion from the collected images; for each extracted object determining an object size, correcting the object size for optical depth perception distortion based upon known or derived geometrical and optical information, for each extracted object determining, an object size, an object aspect ratio, an object speed and object motion path track, comparing the distance corrected object size with a statistical bell curve tuned to nominal human size, comparing aspect ratio to a statistical bell curve tuned to nominal aspect ratio of human proportions including one for walking and one for crawling, comparing object speed to a statistical bell curve tuned to nominal travel speed of a typical human including one for walking and one for crawling and comparing object motion path track against ramp function of absolute distance traveled over time, a method for interactive control to reshape and re-characterize each reference bell curve and motion ramp, determining a respective threat probability for each of the extracted object signatures based upon the corrected object size compared with the bell curve tuned to human size, the aspect ratio compared with the bell curves tuned to both walking and crawling, the object speed compared with the bell curves for walking and crawling and upon sustained motion path over time, summing the respective threat probabilities for each extracted moving object based upon size, aspect ratio, speed and sustained motion path and generating an motion alert upon detecting the summed threat probabilities exceeding a programmable threshold value.

Claims

1. A method of detecting human motion comprising:

a camera collecting video images of a secured area;

extracting objects in motion from the collected images;

for each extracted object determining an object size;

correcting the object size for optical depth perception distortion based upon known or derived geometrical and optical information;

for each extracted object determining, an object size, an object aspect ratio, an object speed and object motion path over time;

extracted a signature for each extracted object by comparing the distance corrected object size with a bell curve tuned to a statistical average of human size, the object aspect ratio based upon a pair of bell curves tuned to a statistical average of aspect ratios of human proportions including one for walking and one for crawling, the object speed with a pair of bell curves tuned to a statistical average of travel speeds of a typical human including one for walking and one for crawling and object motion path travel as measured against absolute distance traveled over time;

determining a respective threat probability for each of the extracted object signatures based upon the corrected object size compared with the bell curve tuned to human size, the aspect ratio compared with the bell curves tuned to both walking and crawling, the object speed compared with the bell curves for walking and crawling and upon a time of sustained object direction of travel;

summing the respective threat probabilities for each extracted object signature based upon size, aspect ratio, speed and time of sustained travel; and

generating a motion alert upon detecting the summed human motion threat probabilities exceeding a programmable threshold value.

2. The method of detecting human motion as in claim 1 wherein the step of correcting for optical depth perception distortion further comprises determining actual and correct distance and height of the extracted object from the camera based upon known or derived geometrical and optical information as mounting height of the camera, optical field of view of the camera, vertical resolution of the camera, mounted angle of the camera relative to camera mounting surface and a horizontal reference parallel to floor and ceiling.

3. The method of detecting human motion as in claim 2 wherein the step of correcting for optical depth perception distortion further comprises the steps in determining actual and correct object distance and height by the calculation sequence of:

camera viewing angle to foot of object, camera viewing angle to head of object, depth to object by the mounting height divided by the tangent of angle to foot, height above object by the tangent of angle to head as multiplied by depth of object and height of object by the difference of camera mounting height and height above object.

4. The method of detecting human motion as in claim 2 wherein the step of correcting for optical depth perception distortion further comprises determining actual and correct vertical and horizontal size of the extracted object by knowing or deriving:

mounting height of the camera, optical field of view of the camera, vertical resolution of the camera, mounted angle of the camera relative to camera mounting surface, a horizontal reference parallel to floor and ceiling and by the calculation sequence of camera viewing angle to foot of object, camera viewing angle to head of object, depth to object by the mounting height divided by the tangent of angle to foot, height above object by the tangent of angle to head as multiplied by depth of object and height of object by the difference of camera mounting height and height above object, determining vertical size equal to object height and horizontal size equal to object height divided by aspect ratio.

5. The method of detecting human motion as in claim 1 wherein the step of determining object threat potential based upon object size further comprises comparing the actual and correct size of a moving object in question with a bell curve tuned to a statistical average of human size, the result of such comparison yielding a linear threat calculation value in direct proportion to nominal human size.

6. The method of detecting human motion as in claim 1 wherein the step of determining object threat potential based upon object aspect ratio further comprises comparing the aspect ratio of a moving object in question with a pair of bell curves tuned to the statistical average of aspect ratios of the human form including one for upright and one for crawling, the result of such comparison yielding a linear threat calculation value in direct proportion to nominal human aspect ratio, as set by the observed moving object orientation.

7. The method of detecting human motion as in claim 1 wherein the step of determining object threat potential based upon object speed further comprises comparing the speed of a moving object in question with a pair of bell curves tuned to the statistical average travel speed of the humans including one for walking and one for crawling, the result of such comparison yielding a linear threat calculation value in direct proportion to nominal human travel speed, as set by the observed moving object orientation.

8. The method of detecting human motion as in claim 1 wherein the step of determining object threat potential based upon object purposeful/intentional travel further comprises tracking and comparing the motion path over time of a moving object in question with a linear ramp function mapping threat potential to absolute distance traveled over time, the result of such comparison yielding a linear threat calculation value in direct proportion to sustained directional movement over time and inversely proportional to oscillatory or wandering movement over time.

9. The method of detecting human motion as in claim 1 wherein the step of determining the respective threat probability for the extracted objects based upon the measurement of object size measured against the bell curve tuned to the statistical average of human size, the extracted object aspect ratio measured against the bell curves tuned to both walking and crawling, the extracted object speed measured against the bell curves for walking and crawling and upon extracted object motion path measured against absolute distance traveled over time;

weighting and summing the respective threat probabilities based upon extracted object signature data obtained within the deployed environment.

10. The method of detecting human motion as in claim 1 wherein the step of determining the overall human motion threat potential of an extracted moving object within a secured area is based upon the combined aggregate summation of a plurality of video camera derived object signature extraction and analysis functions which together build an overall linear threat potential value of the identified object in question.

11. The method of detecting human motion as in claim 1 wherein the step of making an absolute and unqualified determination of a human motion event within the secured area is based upon the comparison of the collective summed linear threat potential value as determined by a plurality of video object signature and analysis functions of the moving object in question with a programmable threshold value, the result of which generates a simple motion/no-motion event signal to a decision making control panel.

12. A system detecting human motion comprising:

a camera that collects video images of a secured area;

a programmed processor that extracts objects in motion from the collected images;

a processor that determines an object size;

a processor that corrects the object size for optical depth perception distortion based upon known or derived geometrical and optical information;

a processor that measures a signature for each extracted object by comparing a distance corrected object size against a bell curve tuned to a statistical average of human size, an object aspect ratio against an a pair of bell curves tuned to a statistical average of aspect ratios of human proportions including one for walking and one for crawling;

an object speed against a pair of bell curves tuned to a statistical average of travel speeds of a typical human including one for walking and one for crawling and object motion path travel as measured against absolute distance traveled over time;

a processor that determines a respective threat probability for each of the extracted object signatures based upon the measurement of object size against the bell curve tuned to human size, the aspect ratio against the bell curves tuned to both walking and crawling, the object speed against the bell curves for walking and crawling and upon a time of sustained direction of travel;

a processor that sums the respective threat probabilities for each extracted object based upon size, aspect ratio, speed and time of sustained travel; and

a processor that generates a motion alert upon detecting that the summed human motion threat probabilities exceed a programmable threshold value.

13. The system as in claim 12 wherein the processor that corrects for optical depth perception distortion further comprises a processor that determines actual and correct distance and height of the extracted object from the camera based upon known or derived geometrical and optical information as mounting height of the camera, optical field of view of the camera, vertical resolution of the camera, mounted angle of the camera relative to camera mounting surface and a horizontal reference parallel to floor and ceiling.

14. The system as in claim 13 wherein the processor corrects for optical depth perception distortion further comprises a processor that determines actual and correct object distance and height by the calculation sequence of:

camera viewing angle to foot of object, camera viewing angle to head of object, depth to object by the mounting height divided by the tangent of angle to foot, height above object by the tangent of angle to head as multiplied by depth of object and height of object by the difference of camera mounting height and height above object.

15. The system as in claim 13 wherein the processor corrects for optical depth perception distortion further comprises a processor that determines actual and correct vertical and horizontal size of the extracted object by knowing or deriving:

mounting height of the camera, optical field of view of the camera, vertical resolution of the camera, mounted angle of the camera relative to camera mounting surface, a horizontal reference parallel to floor and ceiling and by the calculation sequence of camera viewing angle to foot of object, camera viewing angle to head of object, depth to object by the mounting height divided by tangent of angle to foot, height above object by tangent of angle to head as multiplied by depth of object and height of object by the difference of camera mounting height and height above object, determining vertical size equal to object height and horizontal size equal to object height divided by aspect ratio.

16. A system as in claim 12 wherein the processor that determines threat potential based upon object size further comprises a processor that compares the actual and correct size of a moving object in question with a bell curve tuned to a statistical average of human size, the result of such comparison yields a linear threat calculation value in direct proportion to nominal human size.

17. A system as in claim 12 wherein the processor that determines threat potential based upon object size further comprises a processor that compares the aspect ratio of a moving object in question with a pair of bell curves tuned to the statistical average of aspect ratios of the human form including one for upright and one for crawling, the result of such comparison yields a linear threat calculation value in direct proportion to nominal human aspect ratio, as set by the observed moving object orientation.

18. A system as in claim 12 wherein the processor that determines threat potential based upon object size further comprises a processor that compares the speed of a moving object in question with a pair of bell curves tuned to the statistical average travel speed of the humans including one for walking and one for crawling, the result of such comparison yields a linear threat calculation value in direct proportion to nominal human travel speed, as set by the observed moving object orientation.

19. A system as in claim 12 wherein the processor that determines threat potential based upon object size further comprises a processor that determines object threat potential based upon object purposeful/intentional travel further comprising tracking and comparing the motion path over time of a moving object in question with a linear ramp function mapping threat potential to absolute distance traveled over time, the result of such comparison yields a linear threat calculation value in direct proportion to sustained directional movement over time and inversely proportional to oscillatory or wandering movement over time.

20. The system as in claim 12 wherein the processor that determines the respective threat probability for the extracted objects based upon the measurement of object size, the aspect ratio, the object speed against the bell curves for walking and upon a time of sustained direction of travel further comprises a processor that weights the respective threat probability based upon extracted object signature data obtained within the deployed environment.

21. The system as in claim 12 further comprising a processor that determines the overall human threat potential of an extracted moving object within a secured area based upon the combined aggregate summation of a plurality of video camera derived object signature extraction and analysis functions that together build an overall linear threat potential value of the identified object in question.

22. The system as in claim 12 further comprising a processor that makes an absolute and unqualified determination of a human motion event within the secured area based upon the comparison of the collective summation of linear threat potential values as determined by a multitude of video object signature and analysis functions of the moving object in question with a programmable threshold value, resulting in the generation of a simple motion/no-motion event signal to a decision making control panel.