Deep Dynamic-Range Compression of Infrared Video Camera.

The visualization of infrared (IR) spectral videos in the high dynamic range (HDR) is a great challenge for traditional display devices. It is necessary to extend its display range of IR videos for user perception. This research aims at leveraging the deep learning approach to adaptive transform IR videos from the high data bits to the low data bits, which is deep dynamic-range compression (DC). We propose a DC transformer (IRDC-Former) to implement adaptive DC of IR video cameras. The simplicity and feasibility of IRDC-Former are demonstrated by HDR infrared videos in different scenes. The demo video is available at https://v.youku.com/v_show/id_XNTg4MDIyNzQ5Mg==.html.