Method for dynamically setting environmental boundary in image and method for instantly determining human activity

• US 8,634,595 B2
• Filed: 08/11/2011
• Issued: 01/21/2014
• Est. Priority Date: 05/04/2011
• Status: Active Grant

• Current Assignee: National Chiao Tung University (Government of The Republic of China)
• Original Assignee: National Chiao Tung University (Government of The Republic of China)
• Inventors: Song, Kai-Tai, Chen, Wei-Jyun
• Primary Examiner(s): MARIAM, DANIEL G
• Application Number:
  US13/207,782
• Publication Number:
  US 20120281918A1
• Time in Patent Office: 894 Days
• Field of Search: 382/103, 382/107, 382/153, 382/190, 382/199, 382/236, 382/305, 348/143, 348/148, 348/149, 348/158, 386/226
• US Class Current: 382/107
• CPC Class Codes: G06T 2207/10004 Still image; Photographic image
• CPC Class Codes: G06T 2207/20164 Salient point detection; Corner detection
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G06V 40/20 Movements or behaviour, e.g. gesture recognition recognition of facial expressions G06V40/16

Abstract

The present invention provides a method for dynamically setting an environmental boundary in an image and a method for instantly determining human activity according to the method for dynamically setting the environmental boundary. The method for instantly determining human activity includes the steps of retrieving at least an initial environmental image with a predetermined angle, and calculating a boundary setting equation of an object and an environmental boundary in the initial environmental image; retrieving a dynamic environmental image having the object by using a movable platform, and figuring out a new environmental boundary; determining a human image in the dynamic environmental image, recording retention time of the human image, and determining a human posture; and determining human location according to the environmental boundary in the dynamic environmental image and the human image, and instantly determining the human activity.

Claims

1. A method for instantly determining human activity, comprising the steps of:

retrieving a first image of an environment using a camera at a first angle and a first location and figuring out a boundary setting equation of a reference object and a first environmental boundary in the first image of the environment;

retrieving a second image of the environment where the reference object is located using the camera at a second angle and a second location different from the first angle and the first location, respectively, and figuring out a second environmental boundary in the environment based on the boundary setting equation and a position where the reference object is in the second image of the environment;

determining a third image of a person who is in the environment, recording retention time of the person staying in the environment, and determining a posture of the person based on the third image; and

determining a position where the person is located in the environment according to the second environmental boundary and the third image, and determining activities of the person according to the position and the posture of the person, and the retention time,

wherein the method is performed using a processor.

2. The method of claim 1, wherein the environmental boundary equation is figured out based on coordinates, slopes and difference of heights of the reference object in the first and second images of the environment.

3. The method of claim 1, wherein the person in the third image is determined by retrieving a foreground image, accessing a specific area in the foreground image by using pixel projection, accessing a specific point in the specific area via histogram of oriented gradient, and processing the specific point by a support vector machine (SVM).

4. The method of claim 1, wherein the posture of the person is determined by star skeleton technology and Hidden Markov model.

5. The method of claim 1, wherein the steps of determining a position where the person is located in the environment and determining activities of the person are performed by a finite state machine.

6. A method for dynamically setting an environmental boundary in an image, comprising the steps of:

retrieving a first image of an environment using a camera at a first angle and a first location, setting a reference object and a first environmental boundary in the environment, and figuring out a boundary setting equation of the reference object and the first environmental boundary in the initial environmental image;

retrieving a second image of the environment where the reference object is located using the camera at a second angle and a second location different from the first angle and the first location, respectively, and determining a position of the reference object in the second image of the environment; and

figuring out a second environmental boundary in the environment according to the boundary setting equation and the position of the reference object in the second image of the environment,

wherein the method is performed using a processor.

7. The method of claim 6, wherein the environmental boundary equation is figured out based on coordinates, slopes and difference of heights of the reference object in the first and second images of the environment.