基于光谱尺度空间与管道滤波的红外目标检测

为了准确地从复杂干扰背景下检测出真实弱小目标，本文引入视觉显著性，设计了基于快速光谱尺度空间与动态管道滤波的红外目标检测算法。基于真实目标与背景内容之间的整体差异，引入快速光谱尺度空间与阈值分割技术，设计视觉显著性机制，对红外图像完成处理，输出全局显著性映射，以高效过滤干扰背景内容。考虑目标与背景的局部特征差异，构建自适应局部对比度增强机制，对粗检测结果实施处理，获取对应的局部显著性映射，改善视觉显著性区域内目标的对比度。引入高斯差分理论，通过估算每一帧红外图像中的目标像素直径，形成动态管道滤波，充分消除虚警，准确识别出弱小目标。多组实验数据显示：较已有的红外目标检测技术而言，在各种不同的复杂背景干扰下，所提算法呈现出更好的检测能力，拥有更理想的接收机工作特性ROC曲线。

Infrared target detection algorithm based on fast spectral scale space and dynamic pipeline filtering

In order to detect the real dim target from the complex jamming background quickly and accurately, visual saliency is introduced to design an infrared target detection algorithm based on fast spectral scale space and dynamic pipeline filtering. According to the difference of global characteristics between target and background, the visual saliency model is designed by introducing the fast spectral scale space and threshold segmentation technology to complete the rough detection of infrared image for outputting the global saliency map, which can suppress the complex background area effectively. Considering the local feature difference between the target and the background, an adaptive local contrast enhancement mechanism is designed to process the coarse detection results for obtaining the corresponding local saliency map, which can improve the contrast of the target in the visual saliency region. The Gaussian difference theory is introduced to estimate the diameter of the target pixel in each frame of infrared image for forming a dynamic pipeline filter, which can eliminate the false alarm and accurately identify the weak and small targets. Several groups of experimental data show that the proposed algorithm has better detection ability and better Receiver Operating Characteristic(ROC) curve under various complex background interferences compared with the existing infrared target detection technology.