Halo artifacts reduction method for variational based realtime retinex image enhancement
Hiroshi Tsutsui, Yoshikawa Satoshi, Hiroyuki Okuhata, Takao Onoye · Asia-Pacific Signal and Information Processing Association Annual Summit and Conference · 2012
In this paper, we propose a novel halo reduction method for variational based Retinex image enhancement. In variational based Retinex image enhancement, a cost function is designed based on the illumination characteristics. The enhanced image is obtained by extracting the illumination component, which gives minimum cost, from the given input image. Although this approach gives good enhancement quality with less computational cost, a problem that dark regions near edges remain dark after image enhancement, known as halo artifact, still exists. In order to suppress such artifacts effectively, the proposed method adaptively adjusts the parameter of the cost function, which influences the trade-off relation between reducing halo artifacts and preserving image contrast. The proposed method is applicable to an existing realtime Retinex image enhancement hardware implementation.