Anisotropic Diffusion for Details Enhancement in Multiexposure Image Fusion

Figure 4

Two layer decomposition of 1-dimensional signal based on Anisotropic Diffusion after 5 iterations with 𝑘=30,𝜆=1/3 and |𝜂_𝑠|=2 (West & East). The 1-dimensional input signal (𝐼) is decomposed into two main components: a low frequency base layer (B) and a high-frequency detail layer (D). Notice that the edges are preserved in the diffused image (i.e., base layer) and the detail layer yields fine details only.