'''
Cell counting.

'''

import cv2
import cv2.cv
import numpy as np
import matplotlib.pyplot as plt
import math


def detect(img):
    '''
    Do the detection.
    '''
    # create a gray scale version of the image, with as type an unsigned 8bit integer
    img_g = np.zeros((img.shape[0], img.shape[1]), dtype=np.uint8)
    img_g[:, :] = img[:, :, 0]
    cv2.imshow('img', img_g)
    cv2.waitKey(0)

    # 1. Do canny (determine the right parameters) on the gray scale image
    edges = cv2.Canny(img_g,50,110)  # TODO!

    # Show the results of canny
    canny_result = np.copy(img_g)
    canny_result[edges.astype(np.bool)] = 0
    cv2.imshow('img', canny_result)
    cv2.waitKey(0)

    # 2. Do hough transform on the gray scale image
    print "circles"
    circles = cv2.HoughCircles(img_g, cv2.cv.CV_HOUGH_GRADIENT,1,100, param1=80,param2=35,minRadius=0,maxRadius=0)
    print "done"
    circles = circles[0, :, :]

    # Show hough transform result
    print "showing"
    showCircles(img, circles)

    # 3.a Get a feature vector (the average color) for each circle
    nbCircles = circles.shape[0]
    features = np.zeros((nbCircles, 3), dtype=np.int)
    for i in range(nbCircles):
        features[i, :] = getAverageColorInCircle(None, int(circles[i, 0]), int(circles[i, 1]),
                                                 int(circles[i, 2]))  # TODO!

    # 3.b Show the image with the features (just to provide some help with selecting the parameters)
    showCircles(img, circles, [str(features[i, :]) for i in range(nbCircles)])

    # 3.c Remove circles based on the features
    selectedCircles = np.zeros((nbCircles), np.bool)
    for i in range(nbCircles):
        if True:  # TODO
            selectedCircles[i] = 1
    circles = circles[selectedCircles]

    # Show final result
    showCircles(img, circles)
    return circles


def getAverageColorInCircle(img, cx, cy, radius):
    '''
    Get the average color of img inside the circle located at (cx,cy) with radius.
    '''
    maxy, maxx, channels = img.shape
    nbVoxels = 0
    C = np.zeros((3))
    # TODO!
    return C


def showCircles(img, circles, text=None):
    '''
    Show circles on an image.
    @param img:     numpy array
    @param circles: numpy array
                    shape = (nb_circles, 3)
                    contains for each circle: center_x, center_y, radius
    @param text:    optional parameter, list of strings to be plotted in the circles
    '''
    # make a copy of img
    img = np.copy(img)
    # draw the circles
    nbCircles = circles.shape[0]
    for i in range(nbCircles):
        cv2.circle(img, (int(circles[i, 0]), int(circles[i, 1])), int(circles[i, 2]), cv2.cv.CV_RGB(255, 0, 0), 2, 8, 0)
    # draw text
    print "drawing"
    if text != None:
        for i in range(nbCircles):
            cv2.putText(img, text[i], (int(circles[i, 0]), int(circles[i, 1])), cv2.FONT_HERSHEY_SIMPLEX, 0.5,
                        cv2.cv.CV_RGB(0, 0, 255))
    # show the result
    print "imshow"
    cv2.imshow('img', img)
    cv2.waitKey(0)


if __name__ == '__main__':
    # read an image
    img = cv2.imread('normal.jpg')

    # print the dimension of the image
    print img.shape

    # show the image
    cv2.imshow('img', img)
    cv2.waitKey(0)

    # do detection
    circles = detect(img)

    # print result
    print "We counted " + str(circles.shape[0]) + " cells."