from PIL import Image
import numpy as np
import cv2

def region_of_interest(image):
    h, w = image.shape[:2]
    rect = np.array([[(round(w/2),h),(round(w/2),0),(w,0),(w,h)]])
    mask = np.zeros_like(image)
    cv2.fillPoly(mask, rect, 255)
    masked_image = cv2.bitwise_and(image, mask)
    return masked_image

image = 'z1.jpg'

img = cv2.imread(image, 1)
orig_img = img.copy()
gray_img = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)

#bilateralny wiecej detekcji, nie poprawil detekcji znaku na odleglosc

#img = cv2.GaussianBlur(gimg, (5, 5), cv2.BORDER_DEFAULT)
img = cv2.bilateralFilter(gray_img,7, 10, 10, cv2.BORDER_DEFAULT)
img = cv2.medianBlur(img, 5)

img = region_of_interest(img)

# param1=165, param2=45, minRadius=5, maxRadius=80
cimg = cv2.cvtColor(gray_img, cv2.COLOR_GRAY2BGR)

circles = cv2.HoughCircles(img,cv2.HOUGH_GRADIENT, 1, 100, param1=165, param2=50, minRadius=5, maxRadius=80)
circles_round = np.uint16(np.around(circles))

detec_circ = []
padding = 5
for i in circles_round[0,:]:
    # draw the outer circle
    cv2.circle(cimg,(i[0],i[1]),i[2],(0,255,0),2)
    # draw the center of the circle
    cv2.circle(cimg,(i[0],i[1]),2,(0,0,255),3)
    detec_circ.append(orig_img[i[1] - i[2] -padding :i[1] + i[2] + padding, i[0] - i[2] - padding : i[0] + i[2] + padding])

#wyciete obrazy sa zachowane w detec_circ

#cv2.namedWindow('crap',cv2.WINDOW_NORMAL)
#cv2.resizeWindow('crap', 500,600)
#cv2.imshow('crap',cimg)
#print(circles_round)

'''print(circles_round[0,0]) #ignore it
print(circles_round[0,1])
print(circles_round[0,1,0])
print(circles_round[0,1,2])
print(len(circles_round[0]))'''
#cv2.imshow('test',test[5])

for x in range (0, len(detec_circ)):
    #cv2.imshow('detec_circ' + str(x),detec_circ[x])
    img = Image.fromarray(detec_circ[x], 'RGB')
    filename = f'{image.split(".")[0]}-wycinek{x}.jpg'
    img = img.resize((256, 256), Image.ANTIALIAS)
    img.save(filename)
cv2.waitKey(0)
cv2.destroyAllWindows()

img = cv2.imread('z1-wycinek0.jpg', cv2.IMREAD_GRAYSCALE)
img2= cv2.imread('70.png', cv2.IMREAD_GRAYSCALE)
img = cv2.medianBlur(img, 5)
img2 = cv2.medianBlur(img2, 5)
orb = cv2.ORB_create(nfeatures=1500)

#detekcja ORB  zurzywa duzo mocy obliczeniowej, nie nadaje sie do obrazu z kamery
kp1, des1 = orb.detectAndCompute(img, None)
kp2, des2 = orb.detectAndCompute(img2, None)

#sprawdzanie podobienstwa bruteforce
bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True) #hamming dla orb, crosscheck bierze tylko najlepsze podobienstwa dla true
matches = bf.match(des1, des2)
matches = sorted(matches, key=lambda x:x.distance) #sortuje wyniki od najmn do najwieks

for m in matches:
    print(m.distance) #pokazuje jak daleko jestesmy od podobienstwa, im mniej tym lepiej

avg = sum((m.distance for m in matches)) / len(matches)
print(f'Wartosc srednia: {avg}')
imgmatch = cv2.drawMatches(img, kp1, img2, kp2, matches[:], None, flags=2)
#print(len(matches))
#img = cv2.drawKeypoints(img, kp1, None)

cv2.imshow("Image", imgmatch)
cv2.waitKey(0)
cv2.destroyAllWindows()