Real-time Background Subtraction and Shadow Detection

This is an example of real-time background subtraction and shadow detection. 100 background images are used for the scene training. Once the scene training is completed, it is then possible to extract the foreground (object) blobs from real time input image stream.

Green: foreground region
Blue: detected shadow
Red: reflection of the object

Category:

Science & Technology

Tags:

• background
• subtracktion
• shadow
• detection

[pantaneto93]

fine, but how did u get that? my method is:

V(x,y)=background vector at position (x,y)

U(x,y)=pixel vector at same position

V -> v(x,y)= V(x,y) / |V(x,y)|

U -> u(x,y)= u(x,y) / |U(x,y)|

now i calculate the scalar product u*v, this value should be near of 1 if the U(x,y) is a shadow pixel of the background V(x,y) and Us(x,y) (saturation) is more than Vs(x,y)... but sometimes takes some pixels of the objects...

i need how did u get that classification.. thnks

[pantaneto93]

hi, how did u find the shadow pixels?