edge_detection.py

import cv2
import numpy as np

# loading image in gray level
img = cv2.imread('images/M8.jpg', 0)
cv2.imshow('origin gray level', img)

# remove noise
img_rmnoise = cv2.GaussianBlur(img, (3,3), 0)
cv2.imshow('remove the noise', img_rmnoise)

# convolute with proper kernels
laplacian = cv2.Laplacian(img_rmnoise, cv2.CV_64F)
sobelx = cv2.Sobel(img_rmnoise, cv2.CV_64F, 1, 0, ksize=5)  # x
sobely = cv2.Sobel(img_rmnoise, cv2.CV_64F, 0, 1, ksize=5)  # y
sobel = cv2.Sobel(img_rmnoise, cv2.CV_64F, 1, 1, ksize=5)  # y

# show the result by different kernel
cv2.imshow('laplacian', laplacian)
cv2.imshow('sobel x', sobelx)
cv2.imshow('sobel y', sobely)
cv2.imshow('sobel', sobel)

# set trackbar for the canny edge detection
def CannyThreshold(low):
    canny = cv2.Canny(img_rmnoise, low, high*ratio, apertureSize = kernel_size)
    dst = cv2.bitwise_and(canny, canny, mask = img_rmnoise)  # just add some colours to edges from original image
    cv2.imshow('canny', dst)

low = 0
high = 100
ratio = 3
kernel_size = 3
canny = cv2.Canny(img_rmnoise, low, high, apertureSize = kernel_size)
cv2.imshow('canny', canny)
# cv2.namedWindow('canny')
cv2.createTrackbar('Min threshold', 'canny', low, high, CannyThreshold)

cv2.waitKey(0)