A subjective and objective quality assessment of tone-mapped images

With the increasing importance of high dynamic range (HDR) imaging and low availability of HDR displays, the need for efficient tone mapping techniques is very crucial. However the tone mapping operators tend to introduce distortions in the HDR images, thus making it visually unpleasant in normal displays. Subjective evaluation of images is important for rating the tone mapping operators as the users should be able to visualize the complete details present in both the brightly and poorly illuminated regions of the scene. To facilitate a systematic subjective study we have created a database of HDR images tone mapped using popular operators. We conducted a subjective study of the tone mapped images, computed objective scores by using some of the state-of-the-art no-reference low dynamic range image quality assessment algorithms and evaluated their performance. We show that a moderate correlation between objective and subjective scores indicates the need for the consideration of human perception in rating tone mapping operators.