Multi-Exposed Image Fusion Using Multiscale-Surround Switching Map

HDR (High Dynamic Range) image synthesis is a method of synthesizing images of various luminance ranges to expand the dynamic range and reproduce scenes observed with the human eye on a display. Acquiring HDR images requires the use of specialized camera equipment, which is very limited. To overcome these material limitations, HDR image synthesis using various low dynamic range (LDR) images is being researched. In this paper, we use the Retinex algorithm to generate multiple scales from two high-exposure and low-exposure images. Multiscale surround switching maps are utilized to selectively extract desired information from images at each scale. The proposed algorithm synthesizes more detail regions at each scale of the image, thereby reducing noise and halo phenomena in the image, increasing sharpness, and improving color balance compared to existing methods.