coin_counter_test_to_find_good_parameter.py

# Rhys Dunn - 2015
# Learning image vision/OpenCV


# import libraries
import numpy as np
import cv2

COINS_PATH = "./coins"
for i in range(10):
    # load image
    img_path = COINS_PATH+"/coins"+str(i)+".jpg"
    img = cv2.imread(img_path)


    # prep image - blur and convert to grey scale
    grey = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    blurred = cv2.medianBlur(grey, 9, 0)
    blurred = cv2.GaussianBlur(blurred, (3, 3),0)
    blurred = cv2.GaussianBlur(blurred, (11, 11),0)
    blurred = cv2.GaussianBlur(blurred, (7, 7),0)



    # show blurred image and grey scaled image
    cv2.imshow("grey scale", grey)
    cv2.imshow("blurred", blurred)
    cv2.waitKey(0)


    # canny edge detector
    outline = cv2.Canny(blurred, 30, 150)


    # show canny edge detector
    cv2.imshow("The edges", outline)
    cv2.waitKey(0)


    # find the contours
    (cnts, _hierarchy) = cv2.findContours(outline, cv2.RETR_EXTERNAL,
                cv2.CHAIN_APPROX_SIMPLE)


    # draw contours: -1 will draw all contours
    cv2.drawContours(img, cnts, -1, (0, 255, 0), 2)
    cv2.imshow("Result", img)
    cv2.waitKey(0)


    # Print how many coins we found
    print("I found "+str(len(cnts))+" coins in "+img_path)