基于双目立体视觉的图像增强

Image enhancement algorithm based on binocular stereo vision

户外视觉系统采集图像时,容易受到低照度等因素的影响,导致捕获的图像退化,视见度降低。本文提出了一种基于双目立体视觉的图像增强算法。该算法首先通过立体匹配处理求解场景视差图像,然后构建广义双边滤波估计图像照度分量,根据Retinex原理求解图像反射分量,实现图像的增强。实验证明,该算法能够有效地改善图像质量,提高双目视觉系统工作的鲁棒性和可靠性。

Low illumination or other factors reduces the visibility of images captured by outdoor vision system. In this paper,we propose a novel image enhancement algorithm based on binocular stereo vision. The proposed algorithm first solve scene parallax images through stereo matching processing, and then estimate the illumination component by generalized bilateral filter. Finally, it realizes the image enhancement by solving the image reflection component according to the Retinex principle. A comparative study demonstrates that our algorithm can achieve image enhancement, and improve the robustness and reliability of binocular vision system.