A Method to Generate Ghost-Free HDR Images in 360 Degree Cameras with Dual Fish-Eye Lens

Ankit Dhiman, Jayakrishna Alapati, Sankaranarayanan Parameswaran, Eunsunb Ahn

Abstract: 360 degree cameras with fish-eye lenses recently observed increase in popularity due to increasing demand of 360 degree content. However, limitation of conventional lenses to capture entire dynamic range of scene restricts these cameras to produce an image with a high dynamic range (HDR). This is a major problem for ultra-wide angle lenses such as a fish-eye lens because larger field of view tends to have large brightness range. This paper discusses the challenges that occur when existing HDR methods are used as it is on fish-eye images, and proposes a novel method to generate a 360 degree HDR image which is fast and efficient. Secondly, this paper proposes a solution to mitigate the brightness artifact that occurs in the overlapped region of two fish-eye views. We prove the efficacy of the algorithm by comparing the generated HDR output with standard normal HDR and 360 degree HDR datasets created by us.