shape_detection.py

###############################################
# Author: Sahir                               #
# Code:   Detecting Shapes from a Noisy Image #
###############################################

#Import the Libraries
import numpy as np
import cv2
import random

#Reading the noisy image
img = cv2.imread("fuzzy.png",1)

#Displaying to see how it looks
cv2.imshow("Original",img)

#Converting the image to Gray Scale
gray = cv2.cvtColor(img,cv2.COLOR_RGB2GRAY)

#Removing Gaussian Noise
blur = cv2.GaussianBlur(gray, (3,3),0)

#Applying inverse binary due to white background and adapting thresholding for better results
thresh = cv2.adaptiveThreshold(blur, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY_INV, 205, 1)

#Checking to see how it looks
cv2.imshow("Binary",thresh)

#Finding contours with simple retrieval (no hierarchy) and simple/compressed end points
_, contours, _ = cv2.findContours(thresh, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)

#Checking to see how many contours were found
print(len(contours))

#An empty list to store filtered contours
filtered = []

#Looping over all found contours
for c in contours:
	#If it has significant area, add to list
	if cv2.contourArea(c) < 1000:continue
	filtered.append(c)

#Checking the number of filtered contours
print(len(filtered))

#Initialize an equally shaped image
objects = np.zeros([img.shape[0],img.shape[1],3], 'uint8')

#Looping over filtered contours
for c in filtered:
	#Select a random color to draw the contour
	col = (random.randint(0,255), random.randint(0,255), random.randint(0,255))
	#Draw the contour on the image with above color
	cv2.drawContours(objects,[c], -1, col, -1)
	#Fetch contour area
	area = cv2.contourArea(c)
	#Fetch the perimeter
	p = cv2.arcLength(c,True)
	print(area,p)

#Finally show the processed image
cv2.imshow("Contours",objects)
	
#Closing protocol
cv2.waitKey(0)
cv2.destroyAllWindows()