import cv2

import sys

import numpy as np

from sklearn.cluster import MeanShift

from mpl_toolkits.mplot3d import Axes3D

from counter import count

img = cv2.imread('Picture/08-dice.jpg')

img = cv2.resize(img,(500,500))

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

cv2.equalizeHist(gray)

blur = cv2.GaussianBlur(gray, (5, 5),1)

binary = cv2.Canny(blur,70,100)

binary = cv2.dilate(binary,None)

cv2.imshow('canny',binary)

cv2.floodFill(binary,None,(0,0),255)

binary = cv2.dilate(binary,None)

binary = cv2.erode(binary,None,iterations = 2)

binary = cv2.bitwise_not(binary)

dot_mask = np.zeros((binary.shape[0],binary.shape[1]),np.uint8)

num_dot,markers = cv2.connectedComponents(binary)

num_dot -=1

positions = []

for dot in range(num_dot):

index = np.where(markers==dot+1)

pos_x = int((index[0].min()+index[0].max())/2+0.5)

pos_y = int((index[1].min()+index[1].max())/2+0.5)

len_x = int(np.abs(index[1].min()-index[1].max()))

len_y = int(np.abs(index[0].min()-index[0].max()))

radius = np.maximum(len_y,len_x)

positions.append([pos_x,pos_y,radius])

dot_mask[pos_x-int(radius/2):pos_x+int(radius/2),pos_y-int(radius/2):pos_y+int(radius/2)]=255

dot_mask = cv2.erode(dot_mask,None,iterations = 10)

dot_mask = cv2.dilate(dot_mask,None,iterations = 15)

cv2.imshow('this',dot_mask)

num_dot,markers = cv2.connectedComponents(dot_mask)

for dot in range(num_dot-1):

index = np.where(markers==dot+1)

this_mask = np.zeros((markers.shape[0],markers.shape[1],3),np.uint8)

this_mask[markers==dot+1]=255

this_mask = cv2.bitwise_and(this_mask,img)

# cv2.imshow(str(dot),this_mask)

# cv2.waitKey()

dot_img = np.zeros((index[0].max()-index[0].min(),index[1].max()-index[1].min(),3),np.uint8)

dot_img = this_mask[index[0].min():index[0].max(),index[1].min():index[1].max()]

img = count(img,this_mask,index[0].min(),index[1].min())

# cv2.imshow(str(dot),dot_img)

# cv2.waitKey()

print(num_dot)

cv2.imshow('num',img)

cv2.waitKey(0)