System and method for motion detection in a surveillance video

• US 8,571,261 B2
• Filed: 04/22/2010
• Issued: 10/29/2013
• Est. Priority Date: 04/22/2009
• Status: Active Grant

• Current Assignee: CheckVideo, LLC (Kastle Systems International LLC)
• Original Assignee: CheckVideo, LLC (Kastle Systems International LLC)
• Inventors: Gagvani, Nikhil, Gritai, Alexei
• Primary Examiner(s): Park, Chan S
• Assistant Examiner(s): YEH, EUENG NAN
• Application Number:
  US12/765,199
• Publication Number:
  US 20100290710A1
• Time in Patent Office: 1,286 Days
• Field of Search: 382/103, 382/173, 382/217, 340/541, 342/90
• US Class Current: 382/103
• CPC Class Codes: G06F 2218/14 by matching peak patterns
• 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
• CPC Class Codes: G06V 20/49 Segmenting video sequences, i.e. computational techniques such as parsing or cutting the sequence, low-level clustering or determining units such as shots or scenes
• CPC Class Codes: H04N 5/91 Television signal processing therefor

Abstract

In some embodiments, a non-transitory processor-readable medium stores code representing instructions to cause a processor to smooth a current image of a scene to produce a smoothed image and subtract pixel values of a background image of the scene from corresponding pixel values of the smoothed image to produce an altitude difference image. Pixel values of the altitude difference image are weighted to produce a weighted difference image. The weighted difference image is convolved to produce a convoluted difference image. A threshold is applied to each pixel of the convoluted difference image to produce a thresholded difference image. Pixels having a value less than the threshold are removed from the thresholded difference image and classified as background pixels. Foreground pixels are determined based on the thresholded difference image.

Claims

1. A non-transitory processor-readable medium storing code representing instructions to cause a processor to:

smooth a current image of a scene to produce a smoothed image;

subtract pixel values of a background image of the scene from corresponding pixel values of the smoothed image to produce an altitude difference image;

smooth the altitude difference image with a first standard of deviation of pixel values of the altitude difference image to produce a first smoothed altitude difference image;

smooth the altitude difference image with a second standard of deviation of pixel values of the altitude difference image to produce a second smoothed altitude difference image;

multiply pixel values of the first smoothed altitude difference image with pixel values of the second smoothed altitude difference image to produce a weighted difference image;

convolve the weighted difference image to produce a convoluted difference image;

apply a threshold to each pixel of the convoluted difference image to produce a thresholded difference image, pixels having a value less than the threshold being removed from the thresholded difference image and classified as background pixels; and

determine foreground pixels based on the thresholded difference image.

2. The non-transitory processor-readable medium of claim 1, wherein the code representing instructions to cause a processor to smooth the current image includes code representing instructions to cause a processor to smooth the current image using a Gaussian kernel.

3. The non-transitory processor-readable medium of claim 1, wherein the code representing instructions to cause a processor to convolve the weighted difference includes code representing instructions to cause a processor to convolve the weighted difference using a Laplacian of Gaussian kernel.

4. The non-transitory processor-readable medium of claim 1, further comprising code representing instructions to cause a processor to:

update the threshold for pixels classified as background pixels for a predetermined number of successive frames.

5. The non-transitory processor-readable medium of claim 1, wherein the convoluted difference image is a non-binary edge image.

6. The non-transitory processor-readable medium of claim 1, further comprising code representing instructions to cause a processor to:

update a pixel counter associated with each pixel classified as a background pixel; and

update the threshold for each pixel classified as a background pixel having a pixel counter with a value greater than a predetermined counter threshold.