# -*- coding: utf-8 -*-
import cv2
import numpy as np
def hough_detectline(img):
    thetas=np.deg2rad(np.arange(0,180))
    row,cols=img.shape
    diag_len=np.ceil(np.sqrt(row**2+cols**2))
    rhos=np.linspace(-diag_len,diag_len,int(2*diag_len))
    cos_t=np.cos(thetas)
    sin_t=np.sin(thetas)
    num_theta=len(thetas)
    #vote
    vote=np.zeros((int(2*diag_len),num_theta),dtype=np.uint64)
    y_inx,x_inx=np.nonzero(img)
    #vote in hough space
    for i in range(len(x_inx)):
        x=x_inx[i]
        y=y_inx[i]
        for j in range(num_theta):
            rho=round(x*cos_t[j]+y*sin_t[j])+diag_len
            if isinstance(rho,int):
                vote[rho,j]+=1
            else:
                vote[int(rho),j]+=1
    return vote,rhos,thetas
#image = cv2.imread(r'C:\Users\Y\Desktop\input_0.png')
#image_gray=cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)
#image_binary=cv2.Canny(image_gray,150,255)
image = np.zeros((500,500))
image[10:100, 10:100] = np.eye(90)
accumulator, rhos,thetas= hough_detectline(image)
#look for peaks
idx = np.argmax(accumulator)
##下面兩句是尋找投票器最大值所對應的行與列，最大值對應的行就是rho的索引，對應的列就是theta的索引
#可以用這句代替：row,col=np.unravel_index(idx,ccumulator.shape)
#rho=rho[row],theta=theta[col]
rho = rhos[int(idx/accumulator.shape[1])]
theta = thetas[idx % accumulator.shape[1]]
k=-np.cos(theta)/np.sin(theta)
b=rho/np.sin(theta)
x=np.float32(np.arange(1,150,2))
#要在image 上畫必須用float32，要不然會報錯(float不行)
y=np.float32(k*x+b)
cv2.imshow("original image",image),cv2.waitKey(0)
for i in range(len(x)-1):
    cv2.circle(image,(x[i],y[i]),5,(255,0,0),1)
cv2.imshow("hough",image),cv2.waitKey(0)
print ("rho={0:.2f}, theta={1:.0f}".format(rho, np.rad2deg(theta)))