论文部分内容阅读
针对传统高光谱图像矿物识别方法未能充分利用矿物光谱诊断吸收特征与矿物光谱知识、识别过程人为干预多等问题,提出了一种基于光谱知识的高光谱图像自动识别方法.该方法引入了基于光谱吸收特征与波形特征的光谱知识作为自动识别的标准,利用连续统去除操作增强光谱吸收特征,采取基于光谱主次吸收特征的识别决策策略,建立多级约束准则以提高识别精度及避免误识别,通过利用模拟数据进行算法精度评价并应用航空高光谱成像仪AVIRIS(Airborne Visible/Infrared Imaging Spectrometer)数据进行应用分析与验证.结果表明:当图像信噪比大于200时,识别准确率可以达到80.3%,能够得到良好的识别结果以及较高的精度,并实现了基于高光谱图像的矿物自动识别.
Aiming at the problems that traditional hyperspectral image mineral identification method can not fully utilize the diagnostic absorption characteristics of mineral spectrum and mineral spectrum, and human intervention in recognition process, a new method of automatic recognition of hyperspectral image based on spectral knowledge is proposed. Spectral knowledge of spectral absorption characteristics and waveform characteristics is used as the standard of automatic identification. The continuous spectrum removal operation is used to enhance the spectral absorption characteristics. The identification strategy based on primary and secondary absorption characteristics of the spectrum is adopted. A multi-level constraint criterion is established to improve the recognition accuracy and avoid misidentification The accuracy of the algorithm was evaluated by using the simulation data and the application of the AVIRIS (Airborne Visible / Infrared Imaging Spectrometer) data was analyzed and verified.The results show that the recognition accuracy can reach 80.3 when the signal to noise ratio is more than 200 %, Can get good recognition results and high accuracy, and realize the automatic identification of minerals based on hyperspectral images.