基于自构建融合的户外交通图像全天候去雾方法研究

图像复原作为一种有效的去雾手段在户外监控领域发挥着重要的作用。如何全天候的获得高质量的图像是雾天图像复原研究的前沿课题之一。自构造融合方法有着较强的不均匀亮度抵抗性、色彩保真性，展现出取代暗通道理论的巨大潜力。尽管自构造融合方法在处理过程中需要多幅不同曝光度的图像，但其根据优选方法完全可以弥补其不足，还提高了算法的鲁棒性。鉴于图像分割在传统的图像复原领域广泛应用，本项目拟采用模糊侦测、聚焦侦测和侧窗增强进行自构造实现雾天图像的复原，提高复原的质量和算法的鲁棒性，从而促进雾天图像复原方法在户外监控领域的应用。重点研究：1）设计自然状态下雾天图像的明亮区域进行侦测模型；2）构建侧窗分割方法，设计基于亮度映射图侧窗增强方法；3）结合聚焦映射图，设计基于自构造融合的雾天图像复原方法，提高复原的视觉感和鲁棒性。

Image restoration is an effective tool for image dehazing and plays an important role in the field of outdoor monitoring. Especially, how to achieve high-quality images in all-weather conditions is one of the most frontier research topics nowadays. Compared with the dark channel theory, the self-constructing fusion method has better performance in the aspects of immunity to uneven brightness resistance and color fidelity, which has demonstrated the enormous potential to replace the dark channel theory. Although the self-constructing fusion method always needs multiple images with different exposure degrees during processing, this shortage could be overcomed according to the optimization method, which improves the robustness of the algorithm simultaneously. In view of the wide application of image segmentation in the general field of image restoration, this study will explore novel image restoration method based on blur regions detection, focus region detection and side-window enhancement to implement the image dehazing, so as to improve the quality of recovering images and the robustness of the algorithm efficiently. The results of this study would promote the application of foggy image restoration in the field of outdoor monitoring. This study mainly focuses on the following aspects: (1) design of the brightness region detection model for outdoor hazy images (2) constructing the side-window segmentation method and designing side-window enhanced method based on luminance map; (3) designing novel image dehazing algorithm based on self-constructing fusion method combining with focus map, in order to improve the visualization and robustness of image restoration.