convolve

Signature

convolve(field, kernel) → Field

Sum all values in a neighbourhood, multiplied by their weights

Parameters:

• field (Field) – Field to analyse (array / floating point)

• kernel (Kernel) – Neighbourhood to search. The weights must be floating point and will be used to multiply each cell’s value with.

Returns:

New field (array / floating point)

Description

Focal operation summing all values in a neighbourhood, multiplied by their weights.

Formula for computing a result for a single cell \((row, col)\), given an array \(A\) and a 3x3 kernel \(K\):

(1)\[\sum_{i=-1}^{1} \sum_{j=-1}^{1} K[i][j] \times A[row+i][col+j]\]

No-data handling

As long as there is at least one valid value found within the input neighbourhood, a valid value is written to the focal cell in the output field. Only when no such value is found is a no-data value written. The output field is likely to contain less no-data values than the input field.

Example

C

LUE_Field* array = lue_from_gdal(argument_array_pathname);
LUE_Kernel* kernel = lue_read_kernel(argument_kernel_pathname);
LUE_Field* result = lue_convolve(array, kernel);

lue_to_gdal(result, result_array_pathname, NULL);

lue_destruct(array);
lue_destruct(kernel);
lue_destruct(result);

C++

Field const array = from_gdal(argument_array_pathname);
Kernel const kernel = lue::document::read_kernel(argument_kernel_pathname);
Field const result = convolve(array, kernel);

to_gdal(result, result_array_pathname);

PYTHON

array = lfrx.from_gdal(str(argument_array_path))
kernel = ld.read_kernel(argument_kernel_path)
result = lfrx.convolve(array, kernel)

lfrx.to_gdal(result, str(result_array_path))

array	kernel	result

Source code: C, C++, Python

See also

• See focal_sum() for an operation which sums values without multiplying them by weights