一种基于视觉特性的仿生图像增强算法

An Algorithm for Biomimetic Image Enhancement Based on Human Visual Property

常见的基于人类视觉特性的图像增强算法由于是同时完成动态范围压缩和对比度增强,导致增强图像的整体对比度不高、边缘部分效果不佳.通过分析人类视觉系统的全局和局部自适应调节原理及人眼视网膜神经节细胞感受野的传输特性,提出一种仿生图像增强算法.为适应人类视觉系统对光强的主观感觉特性,对图像作全局亮度对数变换;并利用人眼的主观亮度感觉与实际光强的对数呈局部线性关系的特性,采用视网膜神经元感受野三高斯模型来调整亮度图像的局部对比度;最后利用线性变换恢复图像的彩色信息.实验结果表明,该算法的增强效果良好,特别是对于图像边界处,既能很好地增强边缘对比,又可有效地提升区域亮度对比和亮度梯度信息.

Common image enhancements based on human visual property compress the dynamic range of images while enhancing the contrast, so that the whole contrast is not high and the effects on edges are not satisfactory. By analyzing the global and local adaptation of the human visual system and the transfer property of retinal ganglion cells, an algorithm for biomimetic image enhancement is proposed in this paper. Firstly a global logarithm transformation is carried out on the whole image to adapt the sense property of the human visual system; secondly the tri-Gaussian model is adopted to adjust the local contrast based on the local linear relationship between subjective brightness （intensity as perceived by the human visual system） and the logarithm of the light intensity incident on the eye; finally, a linear transformation is used to convert the enhanced image back to a color image. The experimental results show that our proposed algorithm has better performance compared to other methods. In particular, in terms of image border treatment, the proposed algorithm can not only enhance the border contrast, but also efficiently recover the area luminance contrast and luminance gradient information.