Estimation computation

This step aims to calculate a steady shift in both row and column to establish the starting central point of the disparity intervals that are applied in the matching_cost process.

Phase cross correlation method

The phase cross correlation method depends on the frequency domain. It isolates the phase information of cross-correlation from two similar images.

The phase cross correlation algorithm is divided into 4 steps:

• Firstly, we compute the Fourier transform of both the right and left images.

• Secondly, we calculate the cross-correlation between the two Fourier transforms.

• Then, we identify the maximum peak and retrieve a pixel-level shift.

• Finally, for sub-pixel level shifting, we perform an interpolation around this peak.

Note

Currently, only the phase_cross_correlation method is implemented in Pandora2d. We use the phase_cross_correlation function from scikit-image, for further information please see Scikit-image documentation

Configuration and parameters

Warning

Disparities must not be specified in the configuration file if you are using estimation.

Table 5 Parameters

Name	Description	Type	Default value	Available value	Required
estimation_method	estimation measure	string	None	“phase_cross_correlation”	Yes
range_col	Exploration around the initial disparity for columns	int	5	>2, odd number	No
range_row	Exploration around the initial disparity for rows	int	5	>2, odd number	No
sample_factor	Upsampling factor. Images will be registered to within 1 / upsample_factor of a pixel	int	1	>= 1, multiple of 10	No

Example

{
    "input":
    {
        // ...
    },
    "pipeline":
    {
        "estimation":
        {
            "estimation_method": "phase_cross_correlation",
            "range_col": 5,
            "range_row": 5,
            "sample_factor": 20
        }
        // ...
    },
    "output":
    {
        // ...
    }
}

Outputs:

• Showed in log in verbose mode

• Written in the output configuration file

• Stored in the inputs_dataset