entrega.py

import numpy as np
import scipy as sp 
import scipy.ndimage as ndimage
import math
from skimage import exposure,filters
from matplotlib import pyplot as plt
import cv2
import matplotlib.pyplot as plt
from skimage.transform import hough_circle, hough_circle_peaks,hough_ellipse
from skimage.feature import canny
from skimage.draw import circle_perimeter
from skimage.segmentation import active_contour
from skimage.filters import gaussian
import argparse

def resize(img):
	width = 1024
	height = 720
	#####
	return cv2.resize(img,(width,height), interpolation = cv2.INTER_CUBIC)

def rgb2Blue(img):
	b,g,r = cv2.split(img)
	return b

def rgb2Red(img):
	b,g,r = cv2.split(img)
	return r

def rgb2Green(img):
	b,g,r = cv2.split(img)
	return g

def rgb2Gray(img):
	return cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)

def rgb2lab(img):
	return cv2.cvtColor(img,cv2.COLOR_BGR2LAB)

#############preprocess###########
##Image is split on B,G,R channel
##Red channel is isolated
##Smoothing over the red channel is applied
##Sharpening and Equalization to te image are applied
##A morph closing is applied to remove artifacts
##################################
def preprocess(img):
	b,g,r = cv2.split(img)
	gray = rgb2Red(img)
	gray_blur = cv2.GaussianBlur(gray, (5,5), 0)
	gray = cv2.addWeighted(gray, 1.5, gray_blur, -0.5, 0, gray)
	kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(31,31))
	gray = ndimage.grey_closing(gray,structure=kernel)
	gray = cv2.equalizeHist(gray)	
	#gray = cv2.GaussianBlur(gray, (5,5), 0)
	#gray = cv2.medianBlur(gray,9)

	#gray_la= ndimage.laplace(gray)
	#gray = cv2.GaussianBlur(gray,(3,3),0)
	#gray_eq = cv2.equalizeHist(gray)

	return gray

#############getROI##############
##Image is resized
##We take green channel and smooth it
##Opening is done to remove artifacts, in order to preserve only BRIGHTEST elements
##Now we get the most bright pixel position
##We return that position in a 110x110 window
##It is actually a simple way to detect the optic disc, but it works so..
##################################
def getROI(image):
	image_resized = resize(image)
	b,g,r = cv2.split(image_resized)
	g = cv2.GaussianBlur(g,(15,15),0)
	kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(15,15))
	g = ndimage.grey_opening(g,structure=kernel)	
	(minVal, maxVal, minLoc, maxLoc) = cv2.minMaxLoc(g)

	x0 = int(maxLoc[0])-110
	y0 = int(maxLoc[1])-110
	x1 = int(maxLoc[0])+110
	y1 = int(maxLoc[1])+110
	
	return image_resized[y0:y1,x0:x1]


def getValue(img):
	shapeRow = img.shape[0]
	shapeCol = img.shape[1]
	x = 0 
	y = 0
	acu = 0
	maxloc = []
	for i in range(shapeRow):
		for j in range(shapeCol):
			(minVal, maxVal, minLoc, maxLoc) = cv2.minMaxLoc(img[i-15:j-15,i+15:j+15])
			value = maxVal
			if value > acu:
				acu = value 
				maxloc = maxLoc
	return maxloc

def kmeans(img):
	## K-Means
	criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0)
	flags = cv2.KMEANS_RANDOM_CENTERS
	roi = img
	X = roi.reshape((-1, 1))
	X = np.float32(X)
	compactness,labels,centers = cv2.kmeans(X,3,None,criteria,10,flags)

	result = np.choose(labels, centers)
	result.shape = X.shape

	centers = np.uint8(centers)
	res= centers[labels.flatten()]
	res2 = res.reshape((roi.shape))
	return res2


def checkSide(img):
	shapeRow = img.shape[0]
	shapeCol = img.shape[1]
	if cv2.countNonZero(img[:,0:int(shapeCol/2)]) > cv2.countNonZero(img[:,int(shapeCol/2):shapeCol]):
		return True
	else:
		return False

def checkHigh(img):
	shapeRow = img.shape[0]
	shapeCol = img.shape[1]
	if cv2.countNonZero(img[0:int(shapeRow/2),:]) > cv2.countNonZero(img[int(shapeRow/2):shapeRow,:]):
		return True
	else:
		return False

def canny(img,sigma):
	v = np.mean(img)
	sigma = sigma
	lower = int(max(0, (1.0 - sigma) * v))
	upper = int(min(255, (1.0 + sigma) * v))
	edged = cv2.Canny(img, lower, upper)	
	return edged

def hough(edged,limm,limM):
	hough_radii = np.arange(limm, limM, 1)
	hough_res = hough_circle(edged, hough_radii)
	return hough_circle_peaks(hough_res, hough_radii,total_num_peaks=1)

parser = argparse.ArgumentParser(description='Optic disc segmentation.')
parser.add_argument("-f", dest='file', action='store', type=str, help='The image to process.')
args = parser.parse_args()

print(args.accumulate(args.integers))

############################
##Here we start the process
############################

image = cv2.imread(args.file)
roi = getROI(image)
preprocessed_roi = preprocess(roi)

#im2, contours, hierarchy = cv2.findContours(segmented, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
##We get our ROI and then we preprocess it.
##After preprocess, we apply a canny border algorithm
##We dilate the borders in order to make it easy to detect with hough

edged = canny(preprocessed_roi,0.22)
kernel = np.ones((3,3),np.uint8)
edged = cv2.dilate(edged,kernel,iterations=3)
accums, cx, cy, radii = hough(edged,55,80)
for center_y, center_x, radius in zip(cy, cx, radii):
	circy, circx = circle_perimeter(center_y, center_x, radius)
	try:
		roi[circy, circx] = (220, 20, 20)
	except :
		continue
#
init = np.array([circx, circy]).T


cv2.imshow('Pre-Processed Region of Interest '+str(i),preprocessed_roi)
cv2.imshow('Region of Interest '+str(i),roi)

############################

while(1):     
	k = cv2.waitKey(37)
	if k == 37:
		break