CODEWITHSUNDEEP
pythonopencvimage-processing
Zahra
Zahra

Reputation: 391

How to remove small connected objects using OpenCV

I use OpenCV and Python and I want to remove the small connected object from my image.

I have the following binary image as input:

Input binary image

The image is the result of this code:

dilation = cv2.dilate(dst,kernel,iterations = 2)
erosion = cv2.erode(dilation,kernel,iterations = 3)

I want to remove the objects highlighted in red:

enter image description here

How can I achieve this using OpenCV?

Upvotes: 22

Views: 44562

Answers (4)

Soltius
Soltius

Reputation: 2261

Use cv2.connectedComponentsWithStats (doc):

import cv2
import numpy as np

# Start by finding all of the connected components (white blobs in your image).
# 'im' needs to be grayscale and 8bit.
nb_blobs, im_with_separated_blobs, stats, _ = cv2.connectedComponentsWithStats(im)
# im_with_separated_blobs is an image where each detected blob has a different pixel value ranging from 1 to nb_blobs - 1.
# The background pixels have value 0.

im_with_separated_blobs looks like this : enter image description here

# 'stats' (and the silenced output 'centroids') provides information about the blobs. See the docs for more information. 
# Here, we're interested only in the size of the blobs :
sizes = stats[:, cv2.CC_STAT_AREA]
# You can also directly index the column with '-1' instead of 'cv2.CC_STAT_AREA' as it's the last column.

# A small gotcha is that the background is considered as a blob, and so its stats are included in the stats vector at position 0.

# minimum size of particles we want to keep (number of pixels).
# here, it's a fixed value, but you can set it as you want, eg the mean of the sizes or whatever.
min_size = 150  

# create empty output image with will contain only the biggest composents
im_result = np.zeros_like(im_with_separated_blobs)

# for every component in the image, keep it only if it's above min_size.
# we start at 1 to avoid considering the background
for index_blob in range(1, nb_blobs):
    if sizes[index_blob] >= min_size:
        im_result[im_with_separated_blobs == index_blob] = 255

im_result looks like this : enter image description here

Bonus : An alternative way of coding the same process, taking advantage of indexing a numpy array with another array and of the numpy.where function. It's probably more Pythonic insofar that it avoids the for loop, but is also possibly less flexible rearding other filtering conditions (and perhaps less readable) :

_, im_with_separated_blobs, stats, _ = cv2.connectedComponentsWithStats(im)

sizes = stats[:, cv2.CC_STAT_AREA]

im_result = np.where(sizes[im_with_separated_blobs] >= min_size, im, 0)

The key here is to rely on the indexing of a numpy array with a 2D array. sizes[im_with_separated_blobs] is a 2D numpy array with the same shape as im_with_separated_blobs, but where every value has been replaced by sizes[value], i.e. the size of that blob.

Upvotes: 59

Neeraj Madan
Neeraj Madan

Reputation: 1

For isolated or unconnected blobs, try this:

contours, hierarchy = cv2.findContours(edged, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
mask = np.zeros_like(img)
for contour in contours:
    area = cv2.contourArea(contour)
    if area > noise_removal_threshold:
        cv2.fillPoly(mask, [contour], 255)

You can set noise_removal_threshold to whatever you like and make it relative to the largest contour for example or a nominal value like 100 or 25.

Upvotes: 0

Cris Luengo
Cris Luengo

Reputation: 60444

Removing small connected components by area is called area opening. OpenCV does not have this as a function, it can be implemented as shown in other answers. But most other image processing packages will have an area opening function.

For example using scikit-image:

import skimage
import imageio.v3 as iio

img = iio.imread('cQMZm.png')[:,:,0]

out = skimage.morphology.area_opening(img, area_threshold=150, connectivity=2)

For example using DIPlib:

import diplib as dip

out = dip.AreaOpening(img, filterSize=150, connectivity=2)

PS: The DIPlib implementation is noticeably faster. Disclaimer: I'm an author of DIPlib.

Upvotes: 2

Piglet
Piglet

Reputation: 28940

In order to remove objects automatically you need to locate them in the image. From the image you provided I see nothing that distinguishes the 7 highlighted items from others. You have to tell your computer how to recognize objects you don't want. If they look the same, this is not possible.

If you have multiple images where the objects always look like that you could use template matching techniques.

Also the closing operation doesn't make much sense to me.

Upvotes: 1

Related Questions