python - OpenCV中的阈值未检测到我想要获取的完整对象。我怎样才能解决这个问题？

Question

我想从视频提要中检测出鸡蛋，当我尝试对其使用阈值时，它无法获得完整的鸡蛋。

我尝试从此https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_imgproc/py_thresholding/py_thresholding.html应用不同的阈值步骤

将阈值应用于不同的轮廓，结果如下

ret, img = cap.read()
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)

ret,th1 = cv2.threshold(gray,127,255,cv2.THRESH_BINARY)
th2 = cv2.adaptiveThreshold(gray,255,cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY,11,2)
th3 = cv2.adaptiveThreshold(gray,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY,11,2)
ret2,th4 = cv2.threshold(gray,0,255,cv2.THRESH_BINARY+cv2.THRESH_OTSU)

blur = cv2.GaussianBlur(gray,(5,5),0)
ret3,th5 = cv2.threshold(blur,0,255,cv2.THRESH_BINARY+cv2.THRESH_OTSU)

dummy,cnts,hier = cv2.findContours(th1,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)
for c in cnts:
    M = cv2.moments(c)
    if M["m00"] != 0:
        cX = int(M["m10"] / M["m00"])
        cY = int(M["m01"] / M["m00"])
    else:
        cX, cY = 0, 0
    cv2.drawContours(img, [c], -1, (0, 255, 0), 2)
    cv2.circle(img, (cX, cY), 2, (0, 0, 0), -1)
cv2.imshow("Global",th1)
cv2.imshow("Adaptive Mean",th2)
cv2.imshow("Adaptive Gaussian",th3)
cv2.imshow("Otsu's",th4)
cv2.imshow("Otsu's after Blur",th5)

https://imgur.com/a/qgLMkj6

更新:
使用@Martin的答案后，我想出了这个

/image/2EQVM.jpg

通过获取具有最大面积的轮廓。但也有其他轮廓较大的区域。下一个问题是我该怎么做才能过滤掉下面的其他轮廓？我正在考虑确定哪个轮廓是否有角，因为鸡蛋是椭圆形的。另一种方法是裁剪图像，因为鸡蛋仅在图像的上部，但我不知道如何。

码:

dummy,cnts,hier = cv2.findContours(close.astype(np.uint8),cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)
#print (len(cnts))

for c in cnts:
    M = cv2.moments(c)
    area = cv2.contourArea(c)
    print (area)
    if area >46000:
        if M["m00"] != 0:
            cX = int(M["m10"] / M["m00"])
            cY = int(M["m01"] / M["m00"])
        else:
            cX, cY = 0, 0
        cv2.drawContours(img, [c], -1, (0, 255, 0), 2)
        cv2.circle(img, (cX, cY), 2, (0, 0, 0), -1)
cv2.imshow("th5",img)

Answer 1

您没有吃饱鸡蛋的原因是因为阈值太高。你需要降低一点

喜欢:

limit = 100 # possible lower
ret,th1 = cv2.threshold(gray,limit,255,cv2.THRESH_BINARY)

但是，您的问题很大，因为背景(鸡蛋所在的对象)的颜色与鸡蛋的颜色相同。您可能想尝试边缘检测而不是阈值化。

看一下这个:



在播放图像时，我能够得到边缘(仅中途):



码:

import cv2
import numpy as np
img = cv2.imread('eBxV8IA.jpg')
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray,(15,15),0)
lap = cv2.Laplacian(blur,cv2.CV_64F)
blur = cv2.GaussianBlur(lap,(45,45),0)
cv2.imshow("Global",blur)

我能够准确地检测到鸡蛋，但不幸的是，也有很多噪音


码:

import cv2
import numpy as np
img = cv2.imread('eBxV8IA.jpg')
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray,(15,15),0)
lap = cv2.Laplacian(blur,cv2.CV_64F)
blur = cv2.GaussianBlur(lap,(45,45),0)
blur[blur<0]=0
blur = 255.*blur/np.amax(blur)


dummy,cnts,hier = cv2.findContours(blur.astype(np.uint8),cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)
for c in cnts:
    M = cv2.moments(c)
    if M["m00"] != 0:
        cX = int(M["m10"] / M["m00"])
        cY = int(M["m01"] / M["m00"])
    else:
        cX, cY = 0, 0
    cv2.drawContours(img, [c], -1, (0, 255, 0), 2)
    cv2.circle(img, (cX, cY), 2, (0, 0, 0), -1)
cv2.imshow("Global",img)