System and method of detecting moving objects

• US 10,373,015 B2
• Filed: 01/19/2016
• Issued: 08/06/2019
• Est. Priority Date: 01/19/2015
• Status: Active Grant

• Current Assignee: Hanwha Vision Company Limited (Hanwha Corporation)
• Original Assignee: Hanwha Techwin Co., Ltd. (Hanwha Corporation)
• Inventors: Baek, Youngmin, Jung, Jikhan, Kim, Jisung
• Primary Examiner(s): Moyer, Andrew M
• Assistant Examiner(s): Rosario, Dennis
• Application Number:
  US15/000,359
• Publication Number:
  US 20160210759A1
• Time in Patent Office: 1,295 Days
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30232 Surveillance
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation
• CPC Class Codes: G06T 7/254 involving subtraction of images

Abstract

Provided are a system and method of detecting moving objects. The system stores pixel information regarding each of pixels included in frames of the video in a storage, sets a background model comprising at least one background model element, the at least one background model element indicating at least one of a reference visual property and a reference usage time, determines whether the pixels are background pixels or foreground pixels by comparing the pixel information with the at least one background model element, and updating the background model based on a result of the comparing.

Claims

1. A system for detecting a moving object in a video, the system comprising:

at least one memory configured to store pixel information regarding each of pixels included in frames of the video; and

at least one processor configured to implement;

a background model setting unit which sets a background model comprising at least one background model element, the at least one background model element indicating a reference visual property and a usage time;

a background pixel determining unit which determines whether the pixels are background pixels or foreground pixels by comparing pixel information of a first frame with the at least one background model element, and detects a moving object in the first frame based on pixels which are determined to be the foreground pixels according to the comparing; and

a background model updating unit which periodically updates the background model based on a result of the comparing,

wherein the background model updating unit

expands a range of the reference visual property to include a peripheral video property range when the compared pixel information corresponds to the peripheral video property range, the peripheral video property range representing a visual property range that does not belong to and is adjacent to the range of the reference visual property before the range of the reference visual property is expanded;

generates at least one background model element candidate comprising pixel information and usage time information of one of the foreground pixels which has a predetermined visual property; and

when a number of frames in which one of the at least one background model element candidate is used is greater than a predetermined number, updates the one of the at least one background model element candidate to one of the at least one background model element and deletes a background model element that has not been used for a longest time in the background model, and

wherein the background pixel determining unit detects a moving object in a second frame by comparing pixel information of the second frame with the expanded range of the reference visual property.

2. The system of claim 1, wherein the background model updating unit deletes a background model element that has not been used at least for a predetermined time period in the background model.

3. The system of claim 1, wherein the background model updating unit reduces the peripheral video property range included in the expanded range of the reference visual property by a predetermined rate when no pixels corresponding to the peripheral video property range are detected for a predetermined time period.

4. The system of claim 1, wherein the background model updating unit updates the usage time in the at least one background model element according to the compared pixel information.

5. The system of claim 4, wherein the usage time comprises:

a first time element which identifies a most recent frame in which a corresponding background model element is used,

a second time element which indicates a number of sequential frames in which the corresponding background model element is not used, and

a third time element which indicates a number of frames in which the corresponding background model element is used.

6. The system of claim 1, wherein a number of the at least one background model element is greater than one, and

wherein a number of the at least one background model element candidate is smaller than the number of the at least one background model element.

7. The system of claim 1, wherein the reference visual property indicates at least one of brightness, color, and grayscale.

8. A method of detecting a moving object in a video, the method comprising:

storing pixel information regarding each of pixels included in frames of the video in a storage;

setting a background model comprising at least one background model element, the at least one background model element indicating a reference visual property and a usage time;

determining whether the pixels are background pixels or foreground pixels by comparing pixel information of a first frame with the at least one background model element;

detecting a moving object in the first frame based on pixels which are determined to be the foreground pixels according to the comparing;

updating the background model based on a result of the comparing, wherein the updating comprises expanding a range of the reference visual property to include a peripheral video property range when the compared pixel information corresponds to the peripheral video property range, the peripheral video property range representing a visual property range that does not belong to and is adjacent to the range of the reference visual property before the range of the reference visual property is expanded;

generating at least one background model element candidate comprising pixel information and usage time information of one of the foreground pixels which has a predetermined visual property;

when a number of frames in which one of the at least one background model element candidate is used is greater than a predetermined number, updating the one of the at least one background model element candidate to one of the at least one background model element and deleting a background model element that has not been used for a longest time in the background model; and

detects a moving object in a second frame by comparing pixel information of the second frame with the expanded range of the reference visual property.

9. The method of claim 8, wherein the updating comprises deleting a background model element that has not been used at least for a predetermined time period in the background model.

10. The method of claim 8, wherein the updating comprises reducing the peripheral video property range included in the expanded range of the reference visual property by a predetermined rate when no pixels corresponding to the peripheral video property range are detected for a predetermined time period.

11. The method of claim 8, wherein the updating comprises updating the usage time in the at least one background model element according to the compared pixel information.

12. The method of claim 11, wherein the usage time comprises:

a first time element which identifies a most recent frame in which a corresponding background model element is used,

a second time element which indicates a number of sequential frames in which the corresponding background model element is not used, and

a third time element which indicates a number of frames in which the corresponding background model element is used.

13. The method of claim 8, wherein a number of the at least one background model element is greater than one, and

wherein a number of the at least one background model element candidate is smaller than a number of the at least one background model element.

14. The method of claim 8, wherein the reference visual property indicates at least one of brightness, color, and grayscale.