Estimation of gamma-corrected exposure time ratio in multi-exposure images for removal of moving objects

The constructed high-dynamic-range image from merging standard low-dynamic-range images with different camera exposures contains ghost-like artifacts caused by moving objects in the scene. We present a method to utilize the gamma-corrected exposure time ratio between multi-exposure images for removal of moving objects. Between each consecutive image pair in multi-exposure images, the ratio of their exposure times is computed and raised to the power gamma, and this value is used as a cue to detect the pixels corresponding to the moving objects. We propose a method to estimate this ratio from the observed image intensity values, in case the exposure time information or gamma value is unknown. Then the moving objects in multi-exposure images are removed by replacing the intensity values of the detected moving pixels with their expected background values. Experimental results show that the proposed method could remove fast-moving objects from the original multi-exposure images and construct a ghost-free high-dynamic-range image.