An Adaptive Tone Mapping Algorithm Based on Gaussian Filter

A new adaptive Tone Mapping (TM) algorithm based on Gaussian filter is proposed to display the High Dynamic Range (HDR) image on conventional digital display devices. Unlike the conventional luminance mapping function, the proposed algorithm uses the separated two-dimensional Gaussian filter and empirical parameter to obtain better details and speed operation performance. Gaussian filter is mainly used for edge-preserving smoothing. Empirical parameter is introduced to adaptively adjust the overexposed luminance image after mapping. Multiple regression models are utilized to link the empirical parameter to image information which is the mean, logarithm mean and variance of luminance values. Experimental results show that the proposed algorithm retains acceptable image contrast and color information, moreover, outperforms the previous methods on running speed.