OpenCV如何从噪声和误报中清除轮廓

Question

我在Python中找到了标准函数cv2.findContours的图像轮廓。

但是如你所见，中间有一个点，我不能在不打破轮廓线的情况下过滤它。

我怎样才能删除这种假阳性群集？轮廓之外的那些并不重要。

    gray = cv2.cvtColor(self.img, cv2.COLOR_RGB2GRAY)

    _, mask = cv2.threshold(gray, thresh=152, maxval=162, type=cv2.THRESH_BINARY)

    self.mask = cv2.bitwise_and(gray, mask)
    self.contours, hierarchy = cv2.findContours(self.mask, cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)

Answer 1

避免误报的一个好方法是考虑具有最大面积的轮廓：

import imutils #to use sorting functionality 
#To install imutils run: pip install imutils for python 2.x
#Run: pip3 install imutils for python 3.x

cnts = imutils.grab_contours(cnts) #where cnts is the variable in which contours are stored, replace it with your variable name
cnts = sorted(cnts, key = cv2.contourArea, reverse = True)[:10] #sorts contours according to their area from largest to smallest.

largestCont = cnts[0] #store the largest contour

这将删除小点，同时只给出最大的轮廓