Defines how the image is divided into regions for speckle filtering. Whenever two neighboring pixel disparities differ by more than ComponentThreshold disparities, the two pixels are considered as belonging to separate regions. Consequently, each resulting region will not have discontinuities larger or equal to ComponentThreshold in it’s disparity map area.

The smaller this threshold is set, the smaller the resulting disparity regions will be. Thus setting a smaller ComponentThreshold will result in more regions being filtered out, because some regions fall apart and their sizes drop below RegionSize .



An integer specifying the disparity step size, where surfaces should be cut into separate speckle regions.