CODEWITHSUNDEEP
c++imageopencvimage-processing
yucheng
yucheng

Reputation: 43

Matrix logical operation with different number of channels in OpenCV

I have a problem with matrix logical operation. I want to use bitwise_and with image A and image B to get the result image C.

The image data type are all Mat, image A was processed after some functions and it was a binary image with 3 channels. Image B was also a binary image after some processing but just with 1 channel.

Because the channel numbers are different, I get error when do bitwise_and.

How should I do to merge the channel or any else methods to solve this problem?

Here shows image A,B,C:

image A + image B = image C

Upvotes: 4

Views: 1784

Answers (3)

fatihk
fatihk

Reputation: 7919

You can split the three channel matrix and apply bitwise "and operation" separately on each channel:

vector<Mat> channels(3);
// split img:
split(imageA, channels);

bitwise_and(channels[0], imageB, channels[0]);
bitwise_and(channels[1], imageB, channels[1]);
bitwise_and(channels[2], imageB, channels[2]);

Then merge channels to get 3 channel Mat object:

 merge(channels, imageC);

Upvotes: 1

Gurkan Yesilyurt
Gurkan Yesilyurt

Reputation: 2675

I do like the following in Java.

Thresholded channel 1. webcam_image channel 3.

OPENCV 3.0

// convert 3  channel
Imgproc.cvtColor(thresholded, thresholded, Imgproc.COLOR_GRAY2BGR);
Core.bitwise_and(thresholded, webcam_image, webcam_image);

Maybe it'll help you.

Upvotes: 1

Miki
Miki

Reputation: 41765

You need A and B to have same size, type and number of channels. You can use cvtColor to convert from 3 to 1 channel, or viceversa. It depends on the type you want for C:

Mat A; // CV_8UC3
Mat B; // CV_8UC1

If you want C to be 3 channels:

cvtColor(B, B, COLOR_GRAY2BGR);
Mat C;
bitwise_and(A,B,C); // C will be 3 channel, CV_8UC3

else if you want C to be 1 channel:

cvtColor(A, A, COLOR_BGR2GRAY);
Mat C;
bitwise_and(A,B,C); // C will be 1 channel, CV_8UC1

Also, if you don't need to use a mask, you can simply do:

Mat C = A & B;

Upvotes: 3

Related Questions