problem to load 32 bit img

asked 2020-05-13 07:53:06 -0500

NL37 gravatar image

updated 2020-05-13 07:54:23 -0500

Hello,

I have a problem to read a 32-bit image. Actually, I can read the image but I don't understand why, the image is a little bit different.

A snipped of my code : (the image is saved in grayscale)

#include <iostream>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgproc/imgproc.hpp>

int main() {
    int  heigh = 6004 ;
    int  width = 7920 ;
    float* imgDark = new float[heigh*width] ;
    cv::Mat imgDarkcv = cv::imread("/home/nl37/Desktop/0uA_no_binning.tif",2) ;
for(int i = 0 ; i<heigh ; i++)
{
        for(int j =0 ; j<width ; j++)
        {
            imgDark[i*width + j] = imgDarkcv.at<float>(i, j) ;
        }
    }
  cv::Mat *img_cp = new cv::Mat(heigh , width, CV_32FC1, imgDark) ;    
  cv::imwrite("/home/nl37/Desktop/LUT1.tif", *img_cp)  ;
  return 0 ;    
}

Then, with imagej, I calculate the difference between 0uA_no_binning.tif and LUT1. The is not 0. I do not understand my error here. Noticed that I need to convert the image in float*.

Thank you for your help !

edit retag flag offensive close merge delete

Comments

1

What is the difference? I mean it's in the size? Values? Do the images look the same? Is the difference important or very small? (like <1% difference for each pixel)

Anyway, I don't understand why you convert the image like this from float to float...

kbarni gravatar imagekbarni ( 2020-05-13 12:18:58 -0500 )edit

The pixels are different. The size of the image is the the same, and the images look the same. The difference is rather important, and I can't process my image because of the difference.

Actually, I get image from camera in 8-bit. Here, I just need to read a result image using openCV in 32-bit. In my algorithm, I do not use cv::Mat, here I only use openCV to read the image, to validate my algorithm. Is it more clear ?

NL37 gravatar imageNL37 ( 2020-05-14 00:59:46 -0500 )edit

I get image from camera in 8-bit.

so please check imgDarkcv.type() before trying to access it as imgDarkcv.at<float>(i, j) (it's probably NOT float)

also, imwrite() may convert your float imge to uchar internally (not sure about tiff, though)

berak gravatar imageberak ( 2020-05-14 03:51:58 -0500 )edit