OpenCV写入一个浮动图像，要使用哪种转换？

Question

I need to store a float image in OpenCV. Converting it to a CV8U image as suggested by @tomriddle_1234 still stores a black png.

reference.type() = 5
reference.channels() = 1
reference.depth() = 5

How can I convert the image to a 8bit or 16bit so that imwrite can store the image, while maintaining it's float property ie: the stored image is not "washed out colours" due to conversion/loss of precision!

imshow("5t aligned Mean", reference); //Displays the correct image
//reference.convertTo(reference, CV_8U); //Convert image to 8Bit INCORRECT
reference.convertTo(reference, CV_8U, 255.0, 1/255.0); //Correct image
imwrite(subject.c_str(), reference); //Stores a completely black png

Any suggestions are much appreciated!

Answer 1

You can convert to 16bit by multiplying each float pixel by 2^16. Floating point images are stored with values between 0-1

opencv will save 16bit uncompressed in PNG and TIFF with the normal imwrite().

(It will also save them as JPEG although I've had less luck finding things that read 16bit jpeg)

Answer 2

使用CV_NORM_MINMAX在0到255之间转换之前对图像进行归一化