Shadow detection in images

• US 20070110309A1
• Filed: 11/17/2005
• Published: 05/17/2007
• Est. Priority Date: 11/17/2005
• Status: Active Grant

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Ibrahim, Mohamed, Rajagopal, Anupama
• Granted Patent: US 7,639,878 B2
• Time in Patent Office: Days
• US Class Current: 382/171
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation
• CPC Class Codes: G06T 7/254 involving subtraction of images
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06T 7/77 using statistical methods

Abstract

In an embodiment, a novel, scene adaptive, robust shadow detection method has been disclosed. An adaptive division image analysis compares one or more images containing objects and shadows to a reference image that is void of the objects and shadows. Ratios are calculated comparing the frame with shadows and/or objects to the reference frame. Shadows are identified by identifying pixels that fall within a certain range of these ratios. In another embodiment, projection histograms are used to identify the penumbra region of shadows in the image.

Claims

1. A process comprising:

capturing a reference image;

capturing a current image;

calculating reference image to current image object plus shadow ratios (RTOPS);

calculating reference image to current image object ratios (RTO);

calculating reference image to current image shadow ratios (RTS);

calculating an average RTOPS ratio;

calculating an average RTO ratio;

calculating an average RTS ratio;

calculating a first range using said RTOPS, RTO, and RTS ratios; and

identifying a pixel as being under a shadow by

multiplying a lower limit of said first range by said average RTOPS ratio and multiplying an upper limit of said first range by said average RTOPS ratio, thereby giving a second range; and

determining whether said pixel falls within said second range.

2. The process of claim 1, further comprising:

computing a column projection histogram comprising a count of the number of foreground pixels in each column;

computing a row projection histogram comprising a count of the number of foreground pixels in each row; and

identifying penumbra of a shadow by determining locations where said histogram counts are less than a threshold value.

3. The process of claim 1, further comprising:

computing a column projection histogram comprising a count of the number of white pixels in each column;

computing a row projection histogram comprising a count of the number of white pixels in each row; and

identifying penumbra of a shadow by determining locations where said histogram counts are less than a threshold value.

4. The process of claim 1, wherein said RTOPS ratios, said RTO ratios, and said RTS ratios are calculated by comparing the intensity of a pixel in said current image to the intensity of a corresponding pixel in said reference image.

5. The process of claim 1, wherein said first range is calculated by comparing said average RTO ratio and said average RTS ratio to said average RTOPS ratio.

6. The process of claim 1, wherein said reference image comprises a background image from background-foreground learning in a video motion detection algorithm.

7. The process of claim 1, wherein said current image comprises:

an object;

a shadow associated with said object; and

a minimum bounded region comprising coordinates enclosing said object and said shadow.

8. A process comprising:

capturing a reference image;

capturing a current image;

identifying a shadow in said current image using an adaptive image division analysis; and

identifying penumbra of said shadow using a projection histogram analysis.

9. The process of claim 8, wherein said adaptive image division analysis comprises:

calculating reference image to current image object plus shadow ratios (RTOPS);

calculating reference image to current image object ratios (RTO);

calculating reference image to current image shadow ratios (RTS);

calculating an average RTOPS ratio;

calculating an average RTO ratio;

calculating an average RTS ratio;

calculating a first range using said RTOPS, RTO, and RTS ratios; and

identifying a pixel as being under a shadow by

multiplying a lower limit of said first range by said average RTOPS ratio and multiplying an upper limit of said first range by said average RTOPS ratio, thereby giving a second range; and

determining whether said pixel falls within said second range.

10. The process of claim 8, wherein said projection histogram analysis comprises:

computing a column projection histogram comprising a count of the number of foreground pixels in each column;

computing a row projection histogram comprising a count of the number of foreground pixels in each row; and

identifying penumbra of a shadow by determining locations where said histogram counts are less than a threshold value.

11. The process of claim 8, further comprising:

computing a column projection histogram comprising a count of the number of white pixels in each column;

computing a row projection histogram comprising a count of the number of white pixels in each row; and

identifying penumbra of a shadow by determining locations where said histogram counts are less than a threshold value.

12. The process of claim 8, wherein said RTOPS ratios, said RTO ratios, and said RTS ratios are calculated by comparing the intensity of a pixel in said current image to the intensity of a corresponding pixel in said reference image.

13. The process of claim 8, wherein said first range is calculated by comparing said average RTO ratio and said average RTS ratio to said average RTOPS ratio.

14. The process of claim 8, wherein said reference image comprises a background image from background-foreground learning in a video motion detection algorithm.

15. The process of claim 8, wherein said current image comprises:

an object;

a shadow associated with said object; and

a minimum bounded region comprising coordinates enclosing said object and said shadow.

16. A machine readable medium comprising instructions thereon for executing a process comprising:

capturing a reference image;

capturing a current image;

calculating reference image to current image object plus shadow ratios (RTOPS);

calculating reference image to current image object ratios (RTO);

calculating reference image to current image shadow ratios (RTS);

calculating an average RTOPS ratio;

calculating an average RTO ratio;

calculating an average RTS ratio;

calculating a first range using said RTOPS, RTO, and RTS ratios; and

identifying a pixel as being under a shadow by

multiplying a lower limit of said first range by said average RTOPS ratio and multiplying an upper limit of said first range by said average RTOPS ratio, thereby giving a second range; and

determining whether said pixel falls within said second range.

17. The machine readable medium of claim 16, further comprising instructions for:

computing a column projection histogram comprising a count of the number of foreground pixels in each column;

computing a row projection histogram comprising a count of the number of foreground pixels in each row; and

identifying penumbra of a shadow by determining locations where said histogram counts are less than a threshold value.

18. The machine readable medium of claim 16, further comprising instructions for:

computing a column projection histogram comprising a count of the number of white pixels in each column;

computing a row projection histogram comprising a count of the number of white pixels in each row; and

identifying penumbra of a shadow by determining locations where said histogram counts are less than a threshold value.

19. The machine readable medium of claim 16, wherein said RTOPS ratios, said RTO ratios, and said RTS ratios are calculated by comparing the intensity of a pixel in said current image to the intensity of a corresponding pixel in said reference image.

20. The machine readable medium of claim 16,

wherein said first range is calculated by comparing said average RTO ratio and said average RTS ratio to said average RTOPS ratio;

and further

wherein said reference image comprises a background image from background-foreground learning in a video motion detection algorithm.