A High Dynamic Range Imaging Algorithm Based on Moving Objects Processing

High dynamic range (HDR) image can faithfully present real world scenes to be widely utilized in surveillance, remote sensing, and space research fields. This paper proposes an effective scheme to synthesize multiple low dynamic range (LDR) images of a scene with different exposures to a single HDR image, and to manipulate the moving objects appearing in some original LDR images to enhance the quality of the final output HDR image. The proposed algorithm first detects where the moving objects locating in each input LDR images and divides the moving-object-free area of each LDR images into unoverlapped blocks with the same size, then evaluate each block’s average intensity and the statistical values of intensities of whole the LDR images. The proposed gives weight for each block according each block’s average intensity. The final output block is assigned as a weighted sum of the input blocks acquired at different exposures. The experimental results show that the proposed algorithm is a simple and high effective HDR imaging scheme with three advantages: (i) the proposed algorithm is time saving, due to operates completely on intensity plane only, (ii) the ghost artifacts in the resultant HDR image can completely be removed or clearly preserved, (iii) there are no restrictions in the image dynamic color ranges in the proposed algorithm.