Filtering of the disparity map

Theoretical basics

The filtering methods allow to homogenize the disparity maps, those available in pandora are :

median filter.
bilateral filter.
median filter for intervals. See Cost volume confidence for more details.

Note

Invalid pixels are not filtered. If a valid pixel contains an invalid pixel in its filter, the invalid pixel is ignored for the calculation

Configuration and parameters

Name	Description	Type	Default value	Available value	Required
filter_method	Filtering method	str		“median”	Yes
filter_size	Filter’s size	int	3	>=1	No

Name	Description	Type	Default value	Available value	Required
filter_method	Filtering method	str		“bilateral”	Yes
sigma_color	Bilateral filter parameter	float	2.0		No
sigma_space	Bilateral filter parameter	float	6.0		No

Name	Description	Type	Default value	Available value	Required
filter_method	Filtering method	str		“disparity_denoiser”	Yes
filter_size	Bilateral filter parameter	int	11		No
sigma_euclidian	Bilateral filter parameter	float	4.0		No
sigma_color	Bilateral filter parameter	float	100.0		No
sigma_planar	Bilateral filter parameter	float	12.0		No
sigma_grad	Bilateral filter parameter	float	1.5		No

Name	Description	Type	Default value	Available value	Required
filter_method	Filtering method	str		“median_for_intervals”	Yes
filter_size	Filter’s size	int	3	>=1	No
interval_indicator	Indicator for which interval to filter. Ex: If cfg contains a step “cost_volume_confidence.intervals” then interval_indicator should be “intervals”	str	“”		No
regularization	Activate regularization	bool	False	True, False	No
ambiguity_indicator	Indicator for which ambiguity to use during regularization. Ex: If cfg contains a step “cost_volume_confidence.amb” then ambiguity_indicator should be “amb”	str	“”		No
ambiguity_threshold	A pixel is regularized if threshold>ambiguity	float	0.6	>0 and <1	No
ambiguity_kernel_size	Ambiguity kernel size for regularization. See publication for details.	int	5	>=0	No
vertical_depth	Depth for graph regularization. See publication for details.	int	2	>=0	No
quantile_regularization	Quantile used for regularization	float	0.9	>=0 and <=1	No

Example

{
    "input" :
    {
        // ...
    },
    "pipeline" :
    {
        // ...
        "cost_volume_confidence.amb":
        {
            "confidence_method": "ambiguity",
            "eta_max": 0.7,
            "eta_step": 0.01
        },
        "cost_volume_confidence.int":
        {
            "confidence_method": "interval_bounds",
            "regularization": false
        },
        // ...
        "filter":
        {
            "filter_method": "median"
        },
        "filter.int":
        {
            "filter_method": "median_for_intervals",
            "interval_indicator": "int",  // Filtering intervals computed in 'cost_volume_confidence.int'
            "regularization": true,
            "ambiguity_indicator": "amb"  // Using the ambiguity computed above for regularization
        }
        // ...
    }
}