Efficient technique for multi-exposure image fusion

Abstract

We propose a simple and an efficient technique for multi exposure image fusion (MEF). A single digital photograph is insufficient to capture all the details in the scene. When there is a change in radiation across a scene, the images produced by the digital cameras usually suffer from lack of details in the underexposed as well as the over-exposed regions owing to their limited dynamic range. In order to address this limitation, efforts have been put to acquire images that can represent the full dynamic range of the real world, usually referred to as the HDR images. Multi-exposure image fusion (MEF) is considered to be an effective technique to solve this problem. In this technique, a fused image is obtained by integrating the images of the same scene taken under different exposure levels. In this work, a novel multi-exposure image fusion technique is proposed to design an algorithm in the sense to gain most of visual details and vivid colors. Proposed technique blends multiple exposures under the base and detail layer decomposition of input images. Construction of blending weights by using luminance component is then performed to alleviate the problem of loss of visual details. The fused base layer and detail layer are then integrated into the final fused image. Proposed MEF technique requires fewer computational operations, preserves edges and reduced spatial artifacts. Extensive experimental results demonstrates that the proposed exposure fusion technique not only significantly outperforms with other state-of-art methods but is also much faster than competing methods and can achieve better image quality.