基于Sigmoid函数局部视觉适应模型的真实影像再现

为解决图像采集与显示设备之间的动态范围差异,色调映射技术试图建立一种由高到低的动态范围映射关系,可用于一般图像的真实影像再现.在不同亮度适应水平下,人眼能产生不同的响应特性及对比度敏感性,从而同时响应不同明暗的光强.由此建立了一种局部视觉适应的再现算法,采用具有参数控制的Sigmoid函数来模拟视觉适应的S形非线性特点,得到不同局部亮度适应水平下的压缩曲线,因此能较好地协调图像整体对比度、亮度调整与局部增强之间的关系,在增强图像较暗区域的同时,极大程度地保持亮区细节.经主观评价与特征统计参数相结合的方法验证,该算法能有效地实现动态范围压缩,保持图像细节而避免伪像,具有一定的色彩恒定性,且复杂度低,具有很好的实用性.

Tone Reproduction Method by a Local Model of Visual Adaptation Based on Sigmoid Function

In order to solve the dynamic range gap between image acquisition and display devices, tone reproduction algorithms are used to generate a visually similar mapping of input luminance to display luminance, which can be imported to realistic image rendition for ordinary images. The task is similar to visual adaptation processes in human visual system. Under different adaptation levels, human eyes have different contrast sensitivity and adaptation mechanisms to cope with high dynamic range scenes, where both too bright and too dark regions contained. So a realistic image rendition method by a local model of visual adaptation is proposed in this paper. The S-shape nonlinear mapping relationship is simulated by parameters controlled Sigmoid function, and its adaptive compression curves are achieved corresponding to various lightness adaptation levels. The method effectively keeps the balance between the wholly tone and luminance and local contrast, and enhances the visibility in dark region and keeps the detail in too bright region at the same time. Subjective assessment with objective featured statistical values is applied in the paper, it's approved that the method can effectively achieve high dynamic range compression, enhance image detail without artifacts, and it's computationally efficient and easy to use.