Kuo CHEN, Yueting CHEN^*, Huajun FENG, and Zhihai XU

State Key Laboratory of Modern Optical Instrumentation, Zhejiang University, Zhejiang 310027, China

(Received May 23, 2014; Revised August 26, 2014; Accepted September 16, 2014)

Dual sensor cameras are widely used to capture multi-exposure image of high dynamic range scene without ghost effect. The local details and luminance contrast can not be achieved well at the same time by conventional exposure fusion. A novel technique of exposure fusion is proposed to balance the local details and global luminance adaptively for dual sensor camera. Such fusion weight map is calculated by a new down-up-sampling method. And then, a guided filter is employed to refine the weight map and exposure fusion is realized using pixel by pixel approach. Finally, multiple experiments are carried out and six common exposure fusion algorithms are compared to verify the proposed exposure fusion technique. The experimental results show that the proposed method performs excellently and robustly with highest spatial frequency and visual fidelity.

Key words: dual sensor camera, high dynamic range, image fusion, guided filter, weight map

^*E-mail address: chenyt@zju.edu.cn