Gradient field multi-exposure images fusion for high dynamic range image visualization

This paper presents a novel method for fusing multi-exposure images into a low dynamic range (LDR) image that is suitable for display and visualization but it contains details in the high dynamic range (HDR) counterpart. Fused gradient field is derived from the structure tensor of inputs based on multi-dimensional Riemannian geometry with a Euclidean metric assumed. Afterwards, a new method is proposed for modifying the gradient field iteratively with twice average filtering and nonlinearly compressing in multi-scales. These modification operations are all done at the finest resolution. The result is obtained through solving a Poisson equation then linearly stretching to the common range. Experimental results demonstrate the efficiency and effectiveness of this method.

► Fused gradient field is derived from the structure tensor of inputs based on Riemannian geometry metric. ► Gradient field is modified iteratively with twice average filtering and nonlinearly compressing. ► Color components are fused by weighted averaging of inputs then scaled corresponding to luminance intensity. ► Dynamic range compression can be done in the gradient field at full resolution.