Low complexity detail preserving multi-exposure image fusion for images with balanced exposure

Abstract

Multi-exposure image fusion has undergone considerable growth in the last few years, but the challenge still remains to design an algorithm which is efficient in fusion yet low in complexity, such that it can be implemented for real time applications on embedded platforms which have low computational power. A simple, low complexity algorithm for multi-exposure image fusion is proposed in this paper to cater to this need. The algorithm is designed for the image sequences that spread evenly over the exposure range, without any bias towards under or overexposedness. The fusion algorithm assigns weights to pixels of images to be fused, based on the exposure of the input images. The fusion algorithm is free from filtering and transformation, and is carried out on per pixel basis, thus making it highly efficient in terms of computational complexity. The experimental results show that our algorithm produces visually comparable results with some of the commonly used algorithms, which are computationally much more complex. Also, the proposed algorithm is immune to ghosting artifacts caused due to spatial misalignment of the input images.