public interface BackgroundAlgorithmGaussian
Background model in which each pixel is modeled as an independent Guassian distribution. For computational efficiency each band is modeled as having a diagonal covariance matrix with off diagonal terms set to zero, i.e. each band is independent. See [1] for a summary. This is an approximation but according to several papers it doesn't hurt performance much but simplifies computations significantly.
Internally background model is represented by two images; mean and variance, which are stored in
GrayF32 images. This allows for the mean and variance of each pixel to be interpolated,
reducing artifacts along the border of objects.
Tuning Parameters:
[1] Benezeth, Y., Jodoin, P. M., Emile, B., Laurent, H., & Rosenberger, C. (2010). Comparative study of background subtraction algorithms. Journal of Electronic Imaging, 19(3), 033003-033003.
| Modifier and Type | Method and Description |
|---|---|
float |
getInitialVariance()
Returns the initial variance assigned to a pixel
|
float |
getLearnRate()
Returns the learning rate.
|
float |
getMinimumDifference() |
float |
getThreshold() |
void |
setInitialVariance(float initialVariance)
Sets the initial variance assigned to a pixel
|
void |
setLearnRate(float learnRate)
Specifies the learning rate
|
void |
setMinimumDifference(float minimumDifference) |
void |
setThreshold(float threshold) |
float getInitialVariance()
void setInitialVariance(float initialVariance)
initialVariance - initial variancefloat getLearnRate()
void setLearnRate(float learnRate)
learnRate - 0 (slow) to 1 (fast)float getThreshold()
void setThreshold(float threshold)
float getMinimumDifference()
void setMinimumDifference(float minimumDifference)