CODEWITHSUNDEEP
c++opencvthreshold
Ja_cpp
Ja_cpp

Reputation: 2334

Thresholding image in OpenCV of min value

I want to use cv::threshold function because it is well optimized rather than looping through the whole image myself.. however it doesn't have a thresholding to a minimum value. That the options available:

enter image description here

What I want is to set pixels that are smaller than a certain value to that value, example:

image.at<float>(j,i) > 0.1f ? image.at<float>(j,i): 0.1f;

Can I do that without using loops?

I tried that:

image.setTo(0.1, image < 0.1);

But it is saying that:

 error: no match for ‘operator<’ (operand types are ‘const cv::UMat’ and ‘double’)

PS: my images are of type cv::UMat

Upvotes: 1

Views: 1570

Answers (2)

S. Huber
S. Huber

Reputation: 1193

Note that for a number x between 0 and 255 the equivalence max(x, thresh) == 255 - min(255-x, 255-thresh) holds and the results stay within [0, 255].

So you can (i) invert the image, (ii) build the max-threshold by 255-thresh, and (iii) invert again.

Note that mathematically for any constant a it holds that max(x, thresh) == a - min(a-x, a-thresh), e.g., max(x, thresh) == -min(-x, -thresh). The last one is the preferred one for floating-point data.

Upvotes: 1

Ja_cpp
Ja_cpp

Reputation: 2334

I found this function:

void cv::max(
const Mat& src1, // Input array
double value, // Scalar input
Mat& dst // Result array
);

It compares every element of src1 with value and take the maximum, it kind of threshold but with cv::max function.

Upvotes: 2

Related Questions