Navigation College, Dalian Maritime University, Dalian 116026, China.
Key Laboratory of Space Active Opto-Electronic Technologies, Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai 200083, China.
Sensors (Basel). 2022 Sep 29;22(19):7403. doi: 10.3390/s22197403.
Thermal infrared hyperspectral imager is one of the frontier payloads in current hyperspectral remote sensing research. It has broad application prospects in land and ocean temperature inversion, environmental monitoring, and other fields. However, due to the influence of the production process of the infrared focal plane array and the characteristics of the material itself, the infrared focal plane array inevitably has blind pixels, resulting in spectral distortion of the data or even invalid data, which limits the application of thermal infrared hyperspectral data. Most of the current blind pixels detection methods are based on the spatial dimension of the image, that is, processing single-band area images. The push-broom thermal infrared hyperspectral imager works completely different from the conventional area array thermal imager, and only one row of data is obtained per scan. Therefore, the current method cannot be directly applied to blind pixels detection of push-broom thermal infrared hyperspectral imagers. Based on the imaging principle of push-broom thermal infrared hyperspectral imager, we propose a practical blind pixels detection method. The method consists of two stages to detect and repair four common types of blind pixels: dead pixel, dark current pixel, blinking pixel, and noise pixel. In the first stage, dead pixels and dark current pixels with a low spectral response rate are detected by spectral filter detection; noise pixels are detected by spatial noise detection; and dark current pixels with a negative response slope are detected by response slope detection. In the second stage, according to the random appearance of blinking pixels, spectral filter detection is used to detect and repair spectral anomalies caused by blinking pixels line by line. In order to verify the effectiveness of the proposed method, a flight test was carried out, using the Airborne Thermal-infrared Hyperspectral Imaging System (ATHIS), the latest thermal infrared imager in China, for data acquisition. The results show that the method proposed in this paper can accurately detect and repair blind pixel, thus effectively eliminating spectral anomalies and significantly improving image quality.
热红外高光谱成像仪是当前高光谱遥感研究中的前沿有效载荷之一，在陆地和海洋温度反演、环境监测等领域具有广泛的应用前景。然而，由于红外焦平面阵列的生产工艺以及材料本身的特性影响，红外焦平面阵列不可避免地存在盲元，导致数据的光谱变形甚至无效数据，这限制了热红外高光谱数据的应用。目前大多数盲元检测方法都是基于图像的空间维度，即处理单波段区域图像。推扫式热红外高光谱成像仪的工作方式与传统面阵热像仪完全不同，每次扫描仅获得一行数据。因此，目前的方法不能直接应用于推扫式热红外高光谱成像仪的盲元检测。基于推扫式热红外高光谱成像仪的成像原理，我们提出了一种实用的盲元检测方法。该方法包括两个阶段，用于检测和修复四种常见类型的盲元：死元、暗电流元、闪烁元以及噪声元。在第一阶段，通过光谱滤波检测来检测和修复光谱响应率低的死元和暗电流元；通过空间噪声检测来检测噪声元；通过响应斜率检测来检测具有负响应斜率的暗电流元。在第二阶段，根据闪烁元的随机出现，通过光谱滤波检测逐行检测和修复闪烁元引起的光谱异常。为了验证所提出方法的有效性，进行了飞行试验，使用中国最新的机载热红外高光谱成像系统（ATHIS）进行数据采集。结果表明，本文提出的方法能够准确地检测和修复盲元，从而有效地消除光谱异常，显著提高图像质量。
