MINMAX normalization on CV_32F image does not seem to work (all outputs are 0)

C++
1

Hello,

I have a CV_32FC1 image img containing values in the approximate range -1e-04, 1e-03.

I want to visualize this image, so I normalize it as follows

cv::Mat disp;
cv::normalize(img, disp, 0, float_max, cv::NORM_MINMAX, CV_32FC1);

where float_max is the maximum value a float can hold.

When I print to terminal the intensity values of disp they are all 0.
I was expecting the maxima of the image to be near 1e13and the minima to be 0.

Why is that? Am I missing something or is this a bug?

3

Doc is here

A minimal example is welcome. May be you can modify doc example

vector<double> positiveData = { 2.0, 8.0, 10.0 };
vector<double> normalizedData_l1, normalizedData_l2, normalizedData_inf, normalizedData_minmax;
// Norm to probability (total count)
// sum(numbers) = 20.0
// 2.0      0.1     (2.0/20.0)
// 8.0      0.4     (8.0/20.0)
// 10.0     0.5     (10.0/20.0)
normalize(positiveData, normalizedData_l1, 1.0, 0.0, NORM_L1);
// Norm to unit vector: ||positiveData|| = 1.0
// 2.0      0.15
// 8.0      0.62
// 10.0     0.77
normalize(positiveData, normalizedData_l2, 1.0, 0.0, NORM_L2);
// Norm to max element
// 2.0      0.2     (2.0/10.0)
// 8.0      0.8     (8.0/10.0)
// 10.0     1.0     (10.0/10.0)
normalize(positiveData, normalizedData_inf, 1.0, 0.0, NORM_INF);
// Norm to range [0.0;1.0]
// 2.0      0.0     (shift to left border)
// 8.0      0.75    (6.0/8.0)
// 10.0     1.0     (shift to right border)
normalize(positiveData, normalizedData_minmax, 1.0, 0.0, NORM_MINMAX);
4

Hello Laurent, thak you for your answer.

In the end, my problem was that all values in the image were equal, so normalization clipped everything to 0.

Regards

5

No please read doc
double alpha = 1,
double beta = 0,

You set alpha to 0

6

The doc specifies that when normType = NORM_MINMAX, then alpha is the minimum intensity value in dst, while beta is the maximum.

Regards

7

Ok forget everything
I will write a minimal example for you. I think I have to guess your problem…

    Mat img_float = (Mat_<float>(2, 3) << 1.4, 3, 5, -1, -2.2, 0);
    Mat dst;


    cv::normalize(img_float, dst, 0, 1, cv::NORM_MINMAX, CV_32FC1);
    cout << img_float << endl;
    cout << dst << endl;
    cv::normalize(img_float, dst, 0, FLT_MAX, cv::NORM_MINMAX, CV_32FC1);
    cout << img_float << endl;
    cout << dst << endl;

Result

[1.4, 3, 5;
 -1, -2.2, 0]
[0.5, 0.72222221, 1;
 0.16666667, 0, 0.30555555]
[1.4, 3, 5;
 -1, -2.2, 0]
[1.7014118e+38, 2.4575948e+38, inf;
 5.671373e+37, 0, 1.0397517e+38]

Now you are sure that you 've got an inf value that’s fun to process

8

Hello Laurent, thanks for your answer and research.

My problem has been solved since I wrote the second post. Of course float_max was referring to the maximum numerical value a float can hold (not infinity).

As I said in my second post, the problem was that all values in the range were the same, and were therefore squeezed down to zero.

Have a nice day :slight_smile:

9

float_max is not a c++ key word I have to guess it is FLT_MAX