How to write and then read a CV_64FC1 and preserve precision?

nyd
36 ●1 ●4

imread and imwrite apparently don't work with floats, so I should convertto integers, but then, there is no 64bit integer that I could recast to 64bit float. So what do I do?

add a comment

1 answer

Sort by » oldest newest most voted

answered Apr 11 '16

berak
32993 ●7 ●81 ●312

updated Apr 11 '16

if your opencv libs were compiled with openexr support (check cv::getBuildInformation()), you can save the image as .exr or .hdr . this is the only way with imread/imwrite.
you can save it as a text file, using FileStorage:

 Mat floatImg = ...
 FileStorage fs("my.yml", FileStorage::WRITE);
 fs << "myimg" << floatImg;
 fs.release();


 Mat floatImg;
 FileStorage fs("my.yml", FileStorage::READ);
 fs["myimg"] >> floatImg;
 fs.release();