Real-time ghost removal for composing high dynamic range images

It is known that high dynamic range image can be recovered by a set of conventional images shooting at the same scene but with different exposure levels. However, ghost artifacts are commonly seen due to the movement of the object between different captures. In this paper, we propose a real-time algorithm to detect and remove the ghost artifact based on four simple procedures: 1) Inter-pixel Relationship Function estimation for mapping differently exposed images into the same exposure level; 2) bi-directional comparison for further enhancing the detection accuracy, especially at the underexposed and saturated areas; 3) moving object classification with edge artifacts removal; 4) moving object correction for fixing the missing dynamic range. The new algorithm delivers robust ghost removal in O(n) time. To further reduce the processing time, the ghost removal process can be executed in parallel with the photo capturing process.