simpleImageProcessing.py

import math
from scipy import ndimage as ndi
from skimage import feature
import numpy as np 
import os
from scipy import ndimage as ndi
import cv2
import Queue


img = cv2.imread('1.jpeg')
gray = cv2.imrad('1.jpeg', cv2.CV_LOAD_IMAGE_GRAYSCALE)


# Get rid of the tupple



# Gaussian thresholding
img = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 5,2)

# Thresholding Otsu
ret, img = cv2.threshold(img, 0, 255, cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU)



# find all the 'black' shapes in the image
shapeMask = cv2.inRange(img, 200, 255)

# find the contours in the mask

(cnts, _)= cv2.findContours(shapeMask.copy(), cv2.RETR_EXTERNAL,
	cv2.CHAIN_APPROX_SIMPLE)
print "I found %d black shapes" % (len(cnts))
cv2.imshow("Mask", shapeMask)
img = cv2.cvtColor(gray,cv2.COLOR_GRAY2RGB)
# loop over the contours
cv2.waitKey(0)
averageArea = 0
count = 0
list = Queue.Queue()
list_coordY = Queue.Queue()
list_coordX = Queue.Queue()
for c in cnts:
	# draw the contour and show it
	cv2.drawContours(img, [c], -1, (0, 255, 0), 2)
	leftMost = tuple(c[c[:,:,0].argmin()][0])
	rightMost = tuple(c[c[:,:,0].argmax()][0])
	topMost = tuple(c[c[:,:,1].argmin()][0])
	bottomMost = tuple(c[c[:,:,1].argmax()][0])
	y1 = topMost[1]
	y2 = bottomMost[1]
	x1 = leftMost[0]
	x2 = rightMost[0]
	print "\n %d:%d \n", topMost[1] ,bottomMost[1]
	print "\n %d:%d \n", leftMost[0] ,rightMost[0]
	radius1 = bottomMost[1] - topMost[1];
	radius2 = rightMost[0] - leftMost[0];
	crc1 = radius1 * radius1 * math.pi;
	crc2 = radius2 * radius2 * math.pi;
	calArea = cv2.contourArea(c);
	if(crc1 != 0):	
		areaPerError1 = math.fabs((calArea - crc1)/crc1);
		if(areaPerError1 > 0.2):
			print "\n error 1 %d \n", areaPerError1;
	if(crc2 != 0):
		areaPerError2 = math.fabs((calArea - crc2)/crc2);
		if(areaPerError2 > 0.2):
			print "\n error 2 %d \n", areaPerError2;
	
	#cv2.rectangle(img,(bottomMost[0], bottomMost[1]),(topMost[0],topMost[1]),(255,0,0),3)

	
	if(y2 > y1 and x2 > x1):
		crop = img[y1:y2 , x1:x2]
		count += 1
		averageArea += cv2.contourArea(c)
		list.put(crop)
		list_coordY.put((y1,y2)) 
		list_coordX.put((x1,x2)) 
	
cv2.imshow("Image", img)
cv2.waitKey(0)
averageArea = averageArea / count
for i in range(0,list.qsize()):
	crop_img = list.get()
	orig_coordY = list_coordY.get()
	orig_coordX = list_coordX.get()
	x_list = crop_img.shape[0]
	y_list = crop_img.shape[1]
	print "\n reso : %d \n", x_list*y_list
	percentError = ((averageArea - (x_list*y_list))/averageArea)*100
	print "\n percent Error : %d\n", percentError
	if(percentError > 50):
		print "THIS IS CANCELED"
		print "SHAPE : %d %d", crop_img.shape[1], crop_img.shape[0]
		cv2.rectangle(crop_img,(0,0),(crop_img.shape[1], crop_img.shape[0]),(0,0,255),3)
		# Only place corrected images into their spaces
	else:
		list.put(crop_img)
cv2.imshow("Image", img)
cv2.waitKey(0)
cv2.destroyAllWindows()