High dynamic range image composition using a linear interpolation approach

摘要

Linear interpolation is a simple yet effective technique for image composition which works wells for low dynamic range (LDR) images with a fixed range of pixel values. However, it cannot provide good performance for high dynamic range (HDR) images because a high luminance image usually dominates the composition result. This paper proposes a novel algorithm for HDR composition using the linear interpolation. Our scheme decomposes HDR images to be composed into three layers where a linear interpolation can be applied on each layer individually that has been normalized. The algorithm contains three steps including the image decomposition, the image feature composition, and finally the HDR map estimation and image re-rendering. Experimental results show that the proposed approach can produce informative HDR composition images balancing the influence caused by the low or high luminance and preserving the contrast, colors, and salience. The comparison demonstrates that our scheme outperforms the current state-of-the-art methods.