Real Time pixel based Motion Detector - OPENCV

http://sites.google.com/site/ferrariimageprocessing/: This is my pixel based motion detector. It's developed with OpenCV and is based on Kim at al. paper " Real-time foregroundbackground segmentation using codebook model ". Thanks to the caching system it's able to detect forgotten objects.

Category:

Science & Technology

Tags:

• PIXEL
• BASED
• MOTION
• DETECTOR
• KIM
• DAVIS
• OPENCV

[doliharahap]

Hai, is this the same function that codebook in opencv library?

Did you develop this using opencv? if yes, how did you coloring the foreground/the motion object??

[0penCV]

@doliharahap

This example works with my brand new implementation of a codebook model. I use OpenCV just for the basic function (loading images, morphology, drawing, etc..).

[doliharahap]

@0penCV hoo, i see. is your implementation of codebook based on OpenCV library?

[0penCV]

@doliharahap

Yes, I developed a new codebook model starting from the paper I've cited in the description above

[hackish1]

Really interesting.Processing overhead seems to be a frequent problem in these algorithims. How much processing is required to do this? Can it run at 3-4x realtime on a moderatly fast computer?

[0penCV]

@hackish1

I think so. I developed this software on a 1.6 GHz centrico laptop, with no parallel loop. I'm pretty sure with a standard PC (a quad core?) the performance could be improved significantly.

[jesusitotej]

Hello, very good your work. I'm starting to OPENCV2.3 in C + + and maybe you can help and share the code you used for motion detection. From now on I thank you for your generosity. Greetings.

[nu3dvu]

@jonshver

If it may be of value to others, If you are using Internet Explorer 8. The Windows Live tool bar comes with the Bing Translator. I've taken the opportunity to visit Mr. Ferrari's website regarding OpenCV for some time. The Bing Translator has been 90 to 100% accurate in translating his website. once the translator has automatically translated the page one can manually select the 'translated from' language.

[0penCV]

Hi mpillias,

your question is not stupid! If you watch carefully the images, you probably note the blue motion areas contains two-tone blue; the dark tone represent a specific kind of detected motion and the light one another kind of motion. This difference is not significative for pure motion detection, but it's meaningful to detect abandoned object.

[mpillias]

a "stupid" question. i suppose the image on the lower left is showing the absolute difference between subsequent frames. normally this should be black - white image. how do you change to black - blue ?