动态金字塔模型的红外图像SR重建

无人机在复杂飞行过程中，因大气气流及光学设备成像等影响造成采集到的红外图像分辨率过低；另外，因各帧图像分辨率不同，基于固定层数分解的金字塔模型在同一区域下的显著图提取结果存在差异，无法借助视觉技术实现无人机目标定位及自主导航。提出一种改进Itti模型下的红外图像感兴趣区域提取及SR重建算法。算法首先引入多特征对红外图像序列进行金字塔动态分层模型构建；然后，针对不同分辨率下的多帧红外图像进行感兴趣区域的动态提取来克服传统Itti算法的不足；最后，提出基于共轭梯度法的目标函数最小化红外图像超分辨率重建算法，对感兴趣区域进行空间SR重建，提高感兴趣区域目标的空间分辨率。实验验证了提出算法的有效性及准确性。

SR reconstruction algorithm of infrared image based on dynamic pyramid model

During the complex flight process of UAV, the affections such as atmospheric disturbances, optical equipment imaging and other factors, result in low resolution of the acquainted infrared images. In addition, the resolution of each frame infrared image may be different, and extraction of a salient map using the traditional fixed-level decomposition of the pyramid model will be different in the same area, which causes extraction of interested region of UAV difficult, and cannot use visual technology to achieve UAV target positioning and autonomous navigation. In this paper, an improved the interested regional extraction for infrared image and SR reconstruction algorithm based on Itti model was proposed. Firstly, the multi-feature was introduced to construct hierarchical model of the pyramid dynamic of the infrared image sequence. Secondly, the dynamic extraction of the interested region for multi-frame infrared images of different resolution was used to overcome the shortcomings of the traditional Itti algorithm. Finally, the new infrared image reconstruction algorithm based on Fletcher-Reeves majorization-minimization was proposed for spatial SR reconstruction of the interested region to improve the spatial resolution of the interested regional target. Experimental results prove the validity and accuracy of the proposed algorithm.