基于红外视景仿真技术的导向滤波算法

为了实现红外图像去噪声、去模糊的目的，采用综合红外仿真建模技术、导向滤波和盲卷积去模糊的方法，分析了导向滤波算法在视觉图像中的原理，并提出了导向滤波算法的核心——导向图像；采用红外视景仿真技术生成导向图像，解决了导向图像来源问题，并综合导向滤波和盲卷积去模糊算法，有效去除了红外图像噪声和模糊。对综合算法进行了理论分析和实验验证，取得了较满意的试验结果。结果表明，导向滤波算法在处理时间(运行时间降低了25%，从1.1ms降到0.8ms)和运行结果(峰值信噪比从36.7441dB提高到38.5138dB，至少提高了4.5%)上均优于经典滤波算法，在非实时条件下具有很好的处理结果。该研究对红外图像的预处理算法提供了有效的参考。

Guided filtering based on infrared visual simulation

To realize the purpose of infrared image de-noising and de-blurring, infrared simulation modeling, guided filtering and blind convolution methods were studied. The principle of guided filtering in visual was analyzed, and the key idea which is guided image was proposed. In order to conquer guided image in guided filtering method, simulation modeling was used to generate guided image to solve guided image. The guided filter and blind convolution were combined to pre-process infrared image in noise and blur. The theoretical analysis and experimental results were also studied. Result shows that guided filtering is better than the classical algorithm in processing time and result (time reduced 25%, from 1.1ms fall to 0.8ms, and peak signal-to-noise ratio result improve at least 4.5%, from 36.7441dB go up to 38.5138dB). This research provides an effective reference for the infrared image pre-process algorithm.