python-opencv 进行饼乾烤焦检测

这组程序是用来判断饼乾的颜色是否烤焦，不只可以应用在食品业上，在工业上也可以判断金属在热处理後的颜色等等。撷取图片的部分像下面一样，把饼乾撷取出来尽量不要有背景对影像处理比较好，使用HSV的原因是在辨识烤焦时HSV是一个比较powreful的工具，所以使用HSV影像来做直方图比对。(最後下面形态学做比较多次的原因是因为原本这组程序是用在形状比较复杂的饼乾上，可是因为有签保密协定所以不能将图片公布)

import cv2 as cv
import numpy as np
import matplotlib.pyplot as plt
#这个专案是先将图中的饼乾撷取，所以前面做一些形态学把连通区找出...
#将图片找出後再将其转换成HSV进行颜色比对来判断饼乾是否烤焦
#using morphology to find the cookies
#and convert to HSV image to judging cookies is normal or overcook


def eggroll(im,im1):
    global date
    date=cv.absdiff(im, im1)
    return date

def binary(date):
    global binaryim
    ret,binaryim=cv.threshold(date, 10, 255, cv.THRESH_BINARY)
    return binaryim

def close(binaryim):
    global closing
    kernel=np.ones((15, 15),np.uint8)
    clim0=cv.morphologyEx(binaryim,cv.MORPH_CLOSE,kernel)
    closing=cv.morphologyEx(clim0,cv.MORPH_CLOSE,kernel)
    return closing

def open(closing):
    global opening
    kernel=np.ones((15, 15),np.uint8)
    op=cv.morphologyEx(closing,cv.MORPH_OPEN,kernel)
    opening=cv.morphologyEx(op,cv.MORPH_OPEN,kernel)
    return opening

def findedge(img1,img0):
    global conim
    contours,conim=cv.findContours(img1, cv.RETR_TREE, cv.CHAIN_APPROX_SIMPLE)
    c = max(contours, key = cv.contourArea)
    x,y,w,h = cv.boundingRect(c)
    cv.rectangle(img1,(x, y), (x+w, y+h), (0, 255, 0), 2)
    crop_img = im0[y:y+h, x:x+w]
    return crop_img

def findedge2(img3,img4):
    global conim2
    contours1,conim1=cv.findContours(img3, cv.RETR_TREE, cv.CHAIN_APPROX_SIMPLE)
    c1 = max(contours1, key = cv.contourArea)
    x1,y1,w1,h1 = cv.boundingRect(c1)
    cv.rectangle(im4,(x1, y1), (x1+w1, y1+h1), (0, 255, 0), 2)
    crop_img2 = im4[y1:y1+h1, x1:x1+w1]
    return crop_img2

def hsv(im5,im6):
    img2 = cv.cvtColor(im5, cv.COLOR_BGR2HSV)
    img6 = cv.cvtColor(im6, cv.COLOR_BGR2HSV)
    h, s, v = img2[:, :, 0], img2[:, :, 1], img2[:, :, 2]
    hist_h = cv.calcHist([h], [0], None, [256], [0, 256])
    hist_s = cv.calcHist([s], [0], None, [256], [0, 256])
    hist_v = cv.calcHist([v], [0], None, [256], [0, 256])
    h1, s1, v1 = img6[:, :, 0], img6[:, :, 1], img6[:, :, 2]
    hist_h1 = cv.calcHist([h1], [0], None, [256], [0, 256])
    hist_s1 = cv.calcHist([s1], [0], None, [256], [0, 256])
    hist_v1 = cv.calcHist([v1], [0], None, [256], [0, 256])
    match1 = cv.compareHist(hist_h, hist_h1, cv.HISTCMP_BHATTACHARYYA)
    match2 = cv.compareHist(hist_h, hist_h1, cv.HISTCMP_CORREL)
    match3 = cv.compareHist(hist_v, hist_v1, cv.HISTCMP_CHISQR)
    return match2

def judgment(match2):
    date=match2
    if date>=0.95:
        print('excellent')
    elif date>=0.9:
        print('great')
    elif date>=0.85:
        print('normal')
    else:
        print('fail')
        return

im0=cv.imread('dark2.jpg')
im=cv.imread('dark2.jpg',0)
im1=cv.imread('bg.jpg',0)
im3=cv.imread('light3.jpg',0)
im4=cv.imread('light3.jpg')
v=eggroll(im, im1)
v1=binary(v)
v2=close(v1)
v3=open(v2)
v4=findedge(v3, im)
v5=eggroll(im3,im1)
v6=binary(v5)
v7=close(v6)
v8=open(v7)
v9=findedge2(v8,im4)
v10=hsv(v4,v9)
print('correlation',v10)
v11=judgment(v10)
print('judgment',v11)
cv.imshow("cropd",v9)
cv.imshow("cropped", v4)
cv.waitKey(0)