论文部分内容阅读
红外多光谱成像在面对红外干扰和红外隐身时具有全色红外成像手段无法比拟的优势,通过目标在不同波段响应的差异,能有效地克服干扰和检测隐身目标。但传统基于向量的方法在处理多波段图像时没有有效地利用光谱和空间之间的相关性,通过在张量框架下设计辨识方法时,可以综合利用多光谱图像的光谱和空间特性。在设计张量辨识方法时,多光谱图像被作为一个整体来处理,以提高辨识能力。通过改进传统匹配滤波器及有效利用空间和光谱信息,提出了一个Gabor张量匹配滤波模型。该模型充分利用了原有数据的空间、光谱结构,能有效地提高多光谱图像红外目标的辨识能力。
Infrared multispectral imaging has the advantage of panchromatic infrared imaging in the face of infrared interference and infrared stealth, and can effectively overcome the interference and detect stealth targets through the differences of target response in different bands. However, the traditional vector-based method does not effectively utilize the correlation between spectrum and space when dealing with multi-band images. By designing the identification method under the tensor framework, the spectral and spatial characteristics of multi-spectral images can be comprehensively utilized. When designing tensor identification methods, multispectral images are treated as a whole to improve their recognition ability. A Gabor tensor matching filter model is proposed by improving the traditional matched filter and effectively utilizing the spatial and spectral information. The model makes full use of the space and spectral structure of the original data and can effectively improve the recognition ability of multi-spectral image infrared targets.