引导滤波的红外图像预处理算法

Infrared Preprocessing Method Based on Guided Image Filtering

针对红外图像固有噪声和灰度分布非均匀性影响图像质量的问题,提出一种基于引导滤波的红外图像预处理方法。根据引导图的局部信息,在红外图像的不同位置动态生成滤波核函数,从而可以在平滑场景的同时保持甚至增强边缘细节。实验表明,该方法可以有效滤除噪声,保持场景强边缘,改善灰度分布非均匀性;且由于引导滤波算法复杂度与滤波半径无关,处理速度快。此外,应用实验表明该方法可以提高融合图像视觉效果,增强目标跟踪的定位精度,在图像增强中能够有效避免伪边缘的产生。

A new preprocessing scheme for infrared images via guided filtering was proposed. This method gen- erates filter kernel function at different positions in infrared images according to the local information of the guided image. In this way, the processed result can not only get smooth effects but also maintain and even enhance the edge details. The comparing results show that this method can filter noises effectively and meanwhile improve the gray nonuniformity. In addition, our method has a fast speed because the complexity has nothing to do with the filte- ring radius. The application experiments also indicate that this method can improve the visual effects for the fused images, enhance the position accuracy for object tracking and avoid producing false edges in the image enhance- ment.