System and method for low complexity change detection in a sequence of images through background estimation

• US 8,995,718 B2
• Filed: 10/11/2012
• Issued: 03/31/2015
• Est. Priority Date: 10/11/2012
• Status: Active Grant

• Current Assignee: Ittiam Systems Pte Ltd
• Original Assignee: Ittiam Systems Pte Ltd
• Inventors: Sethuraman, Sriram, Chilkundasrinivasamurthy, Santhoshkumar
• Primary Examiner(s): PATEL, KANJIBHAI B
• Application Number:
  US13/649,128
• Publication Number:
  US 20140105498A1
• Time in Patent Office: 901 Days
• Field of Search: 382/103, 382/107, 382/173, 382/180, 382/224, 348/143, 348/154
• US Class Current: 382/103
• CPC Class Codes: G06T 2207/20036 Morphological image processing
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51
• CPC Class Codes: G06T 7/254 involving subtraction of images
• CPC Class Codes: G06V 20/10 Terrestrial scenes scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59

Abstract

A method for low complexity change detection in a sequence of images using configurable block sizes is disclosed. In one embodiment, a first change detection map is generated by performing change detection based on configurable block sizes between a current image and one of an estimated first background image and a previous image. The first change detection map classifies each block as changed or unchanged. The selection between the previous image and the estimated background image for use in change detection is done using a confidence estimate, which is updated both at the low level and at the end of a high-level change analysis. In another embodiment, an estimated second background image is used in addition to the estimate first background image to help quickly adapt when a stationary object starts moving or when a scene object becomes stationary.

Claims

1. A method for low complexity change detection in a sequence of images through background estimation, comprising:

generating a first change detection map by performing change detection based on configurable block sizes between a current image and one of an estimated first background image and a previous image,

wherein the first change detection map classifies each block as changed or unchanged,

wherein generating the first change detection map comprises;

determining a first block level change metric between the current image and one of the estimated first background image and the previous image;

computing a confidence metric based on the determined first block level change metric;

classifying whether each block has changed or unchanged by comparing the associated first block level change metric against an associated block level noise threshold; and

selection of one of the estimated first background image and the previous image based on the confidence metric for generation of the first change detection map for a next image;

identifying each connected region based on the changed blocks and assigning a label to each connected region; and

classifying each labeled region in the current image as changed or unchanged.

2. The method of claim 1, wherein identifying each connected region based on the changed blocks, comprises:

connecting changed blocks by performing morphological filtering on the first change detection map; and

identifying each connected region based on the connected changed blocks.

3. The method of claim 2, wherein identifying each connected region based on the connected changed blocks and assigning the label to each connected region comprises:

grouping the connected changed blocks in a current row of the first change detection map in a raster scan order; and

merging and labeling the groups in the current row with connected groups in a previous row until all the connected groups are labeled.

4. The method of claim 2, wherein classifying each labeled region in the current image as changed or unchanged comprises:

determining a matching labeled region in the previous image for each labeled region in the current image;

computing geometrical properties for each labeled region in the current image; and

classifying each labeled region in the current image as changed or unchanged based on the computed geometrical properties of the matched labeled regions in the current image and the previous image.

5. The method of claim 4, further comprising:

modifying the confidence metric of all blocks of the labeled regions in the current image based on the classification.

6. The method of claim 5, further comprising:

updating the first background image based on the modified confidence metric.

7. The method of claim 6, wherein selection of one of the estimated first background image and the previous image for the next image comprises:

if selection of one of the estimated first background image and the previous image for the current image is equal to the previous image and the confidence metric is greater than a first threshold, then the updated first background image is selected for the next image;

if selection of one of the estimated first background image and the previous image for the current image is equal to the previous image and the confidence metric is less than or equal to the first threshold, then the current mage is selected for the next image;

if selection of one of the estimated first background image and the previous image for the current image is not equal to the previous image and the confidence metric is less than or equal to a second threshold, then the current image is selected for the next image, wherein the second threshold is less than the first threshold; and

if selection of one of the estimated first background image and the previous image for the current image is not equal to the previous image and the confidence metric is greater than the second threshold, then the updated first background image is selected for the next image.

8. The method of claim 1, further comprising:

updating the block level noise threshold and the confidence metric based on the first block level change metric.

9. The method of claim 1, further comprising:

generating a second change detection map by performing change detection based on configurable block sizes between the current image and an estimated second background image, wherein the estimated second background image is updated relatively slower than the estimated first background image, and wherein the second change detection map classifies each block as changed or unchanged.

10. The method of claim 9, wherein generating the second change detection map by performing change detection based on configurable block sizes between the current image and the estimated second background image comprises:

determining a second block level change metric between the current image and the estimated second background image; and

classifying whether each block has changed or unchanged by comparing the associated second block level change metric against the associated block level noise threshold.

11. The method of claim 9, further comprising:

updating the first background image and the second background image based on the classification of each block as changed or unchanged against the estimated first background image and estimated second background image.

12. The method of claim 11, wherein updating the first background image and the second background image comprises:

updating the first background image when the block is classified as unchanged in at least one of the first change detection map and the second change detection map; and

updating both the first background image and the second background image when the block is classified as unchanged in both the first change detection map and the second change detection map.

13. A system for low complexity change detection in a sequence of images through background estimation, comprising:

a motion detector to generate a first change detection map by performing change detection based on configurable block sizes between a current image and one of an estimated first background image and a previous image,

wherein the first change detection map classifies each block as changed or unchanged,

wherein the motion detector is further configured to;

determine a first block level change metric between the current image and one of the estimated first background image and the previous image;

compute a confidence metric based on the determined first block level change metric;

classify whether each block has changed or unchanged by comparing the associated first block level change metric against an associated block level noise threshold; and

select one of the estimated first background image and the previous image based on the confidence metric for generation of the first change detection map for a next image;

a connected component analysis unit to identify each connected region based on the changed blocks and assign a label to each connected region; and

a blob matching unit to classify each labeled region in the current image as changed or unchanged.

14. The system of claim 13, further comprising:

a morphological filter to perform morphological filtering on the first change detection map; and

wherein the connected component analysis unit to;

connect changed blocks based on the morphological filtering on the first change detection map; and

identify each connected region based on he connected changed blocks.

15. The system of claim 14, wherein the connected component analysis unit groups the connected changed blocks in a current roe of the first change detection snap in a raster scan order, and merges and labels the groups in the current row with connected groups in a previous row until all the connected groups are labeled.

16. The system of claim 14, wherein the blob matching unit to:

determine a matching labeled region in the previous image for each labeled region in the current image;

compute geometrical properties for each labeled region in the current image;

classify each labeled region in the current image as changed or unchanged based on the computed geometrical properties of the matched labeled regions in the current image and the previous image; and

modifies the confidence metric of all blocks of the labeled regions in the current image based on the classification.

17. The system of claim 16, further comprising an updating unit to update the first background image based on the modified confidence metric.

18. The system of claim 13, further comprising an updating unit to update the block level noise threshold and the confidence metric based on the first block level change metric.

19. The system of claim 13, wherein the motion detector generates a second change detection map by performing change detection based on configurable block sizes (m×

n pixels) between the current image and an estimated second background image, wherein the estimated second background image is updated relatively slower than the estimated first background image, and wherein the second change detection map classifies each block as changed or unchanged.

20. The system of claim 19, wherein the motion detector determines a second block level change metric between the current image and the estimated second background image, and classifies whether each block has changed or unchanged by comparing the associated second block level change metric against the associated block level noise threshold.

21. The system of claim 20, further comprising an updating unit to update the first background image and the second background image based on the classification of each block as changed or unchanged against the estimated first background image and estimated second background image, and wherein the updating unit updates the first background image when the block is classified as unchanged in at least one of the first change detection map and the second change detection map, and wherein the updating unit updates both the first background image and the second background image when the block is classified as unchanged in both the first change detection map and the second change detection map.

22. An non-transitory computer-readable storage medium for low complexity change detection in a sequence of images through background estimation having instructions that, when executed by a computing device, cause the computing device to perform a method comprising:

generating a first change detection map by performing change detection based on configurable block sizes between a current image and one of an estimated first background image and a previous image,

wherein the first change detection map classifies each block as changed or unchanged,

wherein generating the first change detection map comprises;

determining a first block level change metric between the current image and one of the estimated first background image and the previous image;

computing a confidence metric based on the determined first block level change metric;

classifying whether each block has changed or unchanged by comparing the associated first block level change metric against an associated block level noise threshold; and

selection of one of the estimated first background image and the previous image based on the confidence metric for generation of the first change detection map for a next image;

identifying each connected region based on the changed blocks and assigning a label to each connected region; and

classifying each labeled region in the current image as changed or unchanged.

23. The non-transitory computer-readable storage medium of claim 22, further comprising:

generating a second change detection map by performing change detection based on configurable block sizes between the current image and an estimated second background image, wherein the estimated second background image is updated relatively slower than the estimated first background image, and wherein the second change detection map classifies each block as changed or unchanged.

24. The non-transitory computer-readable storage medium of claim 23, further comprising:

updating the first background image when the block is classified as unchanged in at least one of the first change detection map and the second change detection map; and

updating both the first background image and the second background image when the block is classified as unchanged in both the first change detection map and the second change detection map.