morphomath 0.0.1

fast morphology using kernel decomposition

Useful links: Binary Installers | Source Repository | Online Documentation |

Description

This module enables efficient morphological erosion and dilatation. It uses the kernel subdivision algorithm implemented in C, with multithreading.

Features

1. Works for any tensor dimension, 2d for images, 3d for videos…

2. The morphological structuring element decomposition logarithmically reduces temporal complexity.

3. Functions can be parallelized to take advantage of all the CPU threads, in exchange of higher edge effects.

4. Functions can be compiled dynamically in C to reduce side-effects and overhead, in exchange for a longer loading time.

Examples

import morphomath, cv2
img = cv2.imread("image.png", cv2.IMREAD_GRAYSCALE)
dilate = morphomath.Dilatation([[0, 1, 0], [0, 1, 0], [1, 1, 1]]).decomposed()
print(dilate)
cv2.imwrite("result.png", dilate(img))

Before

After