OpenCV 3 Computer Vision with Python Cookbook

How it works

First, you need to have a set of images with different known exposure times. Modern cameras store a lot of information including exposure time in image files, so it's worth checking the image's properties.

Computing an HDR image starts with recovering CRF (Camera Response Function), which is a mapping between real intensity and the pixel's intensity (which is in the [0, 255] range) for each color channel. Usually it's non-linear, and makes it impossible to simply combine images with different exposures. It can be done by creating an instance of the calibrate algorithm with cv2.createCalibrateDebevec. When the calibration instance is created, you need to invoke its process method and pass an array of images and an array of exposure ...