segment_image

Abstract

• A lib can be used to segment image https://pypi.org/project/segment-image/
• The first version just has k_means algorithm.

install

• python >= 3.6
• install

$ pip install segment-image

how to use

• cv2need to be installed by pip install opencv-python

• Kmeans

import segment_image
import cv2

img = cv2.imread("./sample_imgs/lena.png")
k = 2 # number of segments
k_means = segment_image.Kmeans(img, k)
iteration = 10
convergence_radius = 1e-6
k_means.run(iteration, convergence_radius)

• gvf_snake (The speed is so slow, so I use small max_iteration_gvf and max_iteration_gvf. if you want to get great results, pleas adjust them)

import segment_image
import cv2

img = cv2.imread("./sample_imgs/star.png")
max_iteration_gvf = 10
max_iteration_snake = 10
gvf_snake = segment_image.GVFSnake(img, max_iteration_gvf, max_iteration_snake)
gvf_snake.run(save=True)

example

• https://github.com/MyDuan/segment_image/blob/main/examples/example_of_kmeans.py

• https://github.com/MyDuan/segment_image/blob/main/examples/example_of_gvf_snake.py

• results:

  • kmeans:

  

  • gvf_snake