Search code examples
pythonopencvcontour

Cannot find contours using OpenCV3 but code works on on OpenCV2

Here is the code that does NOT work:

original_image = cv2.imread(test_image,cv2.IMREAD_GRAYSCALE);
bin_ada = cv2.adaptiveThreshold(original_image,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,43,0);
img,contours, hierachy = cv2.findContours(bin_ada,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE,(0,0));

Here is the code that does work:

original_image = cv2.imread(test_image,cv2.IMREAD_GRAYSCALE);
bin_ada = cv2.adaptiveThreshold(original_image,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,43,0);
contours, hierachy = cv2.findContours(bin_ada,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE,(0,0));

The difference is that the first code is running with Python3 and Newer OpenCV libraries (3.4.1) while the second code is running with Python2 and Older OpenCV libraries (2.4.5).

Is this a bug? Or is there a simple explanation as to why the first piece of code does not work?

The error is this:

OpenCV(3.4.1) /io/opencv/modules/imgproc/src/contours.cpp:1894: error: (-215) _contours.empty() || (_contours.channels() == 2 && _contours.depth() == 4) in function findContours

I'd appreciate any help

Solution

  • You don't need the last argument of findContours, it is an offset and you are setting to (0,0) (if you do need an offset use offset=(y,x)).

    This code worked in python 3.6.3 and opencv 3.4

    import cv2
import numpy as np

original_image = cv2.imread("./1.jpeg",cv2.IMREAD_GRAYSCALE);
bin_ada = cv2.adaptiveThreshold(original_image,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,43,0);
img,contours, hierachy = cv2.findContours(bin_ada,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE);