基于mean shift的光晕抑制算法

Halo inhibition algorithm based on mean shift filter

针对单尺度Retinex(single scale retinex,SSR)算法在图像边缘附近容易产生光晕现象,提出了一种基于均值漂移(mean shift,MS)算法的自适应滤波光照估计算法。该算法根据MS算法得到的先验知识,在图像边缘附近自适应地改变滤波器的形状,从而达到抑制光晕的目的。实验表明该算法不仅能够有效地提高图像的对比度,而且相比于其他算法,该算法收敛速度较快,图像主观效果较好。

Since the single scale Retinex（SSR） algorithm is suffered from halo artifacts which are introduced by image edges,an illumination estimation algorithm is proposed to inhibit halo artifacts which is based on mean shift（MS） algorithm and self-adaptive filter.According to the prior knowledge which is obtained by MS algorithm,this algorithm alters the shape of the self-adaptive filter to restrain halo artifacts at the boundaries.Experimental results demonstrate that this algorithm improves the contrast of images,and it also provids with faster convergence and better visual effect than previous algorithms.