主要论述了一种基于引导滤波器图像分层的高动态范围红外图像增强算法，通过引导滤波器将原始红外图像分解成基本层和细节层，再对它们分别采用相应的γ 变换进行压缩，最后按照一定比例将两部分重新合成，从而在保留图像细节的同时有效地使红外场景得到高动态灰度显示。引导滤波器不依赖于滤波半径与图像灰度值范围，执行效率更高，计算速度更快；另外，引导滤波器是局部线性模型，边缘保持特性更好，能克服其他滤波器在图像灰度变化比较剧烈的边缘易出现梯度翻转，造成图像出现“伪边缘”的缺陷。实验结果表明，通过本算法增强后的红外图像，不论是人眼的主观评价还是客观评价，都具有较强的细节增强能力和较佳的视觉表现，且具有实时处理的前景。

This paper analyzes a new high dynamic range infrared image enhancement algorithm, based on a guided image filter with hierarchical techniques. This algorithm adopts a base component and a detail component extracted by the guided image filter, and the two components are compressed through gamma transform to fit the dynamic display range, and then recombined to obtain the output enhancement image. Independence from the filter radius and the image gray value range increases the execution efficiency and calculation speed of the guided filter technique. In addition, guided filter is a local linear model with better edge-keeping, which allows it to avoid in curring gradient flip and causing “pseudo edge” on the edge of the image, where the gray-level change is dramatic, unlike other filters. The experiments described in this paper show that image enhancement using this algorithm results in strong detail enhancement and better visual performance, regardless of subjective(human eyes) or objective evaluation. Moreover, this algorithm also has the potential for use in real-time processing.