Abstract
In this chapter we are presenting the literature and proposed approaches for anomaly detection in hyperspectral images. These approaches are grouped into four categories based on the underlying techniques used to achieve the detection: 1) the statistical based methods, 2) the kernel based methods, 3) the feature selection based methods and 4) the segmentation based methods. Since the first approaches are mostly based on statistics, the recent works tend to be more geometrical or topological especially with high resolution images where the high resolution implies the presence of many materials in the same geographic area that cannot be easily distinguished by usual statistical methods.
TopAnomaly Detection In Hyperspectral Imagery
Problems
The magnitude of anomalies motivated researches in anomaly detection and interpretation for hyperspectral images. Since the beginning, researchers had to cope with the problem of the absence of any prior knowledge about the treated data. Therefore, they try to use statistical methods to compare between the Pixel Under Test (PUT) and the background. For statistical methods the background is modeled with a linear distribution of the Probability Density Function (PDF) that supposes its homogeneity. This supposition accentuates the False Alarm Rate (FAR) especially for high resolution images where the supposition of homogeneity seems to be inappropriate since the big diversity of existing materials. To decrease this fact, non linear models of the background have been proposed with the kernel based anomaly detectors. Other researches try to solve the anomaly detection problem with different techniques as feature selection and the segmentation.
Whatsoever the underlying techniques are, there are three principal challenges to overcome. The first challenge concerns increasing the detection rate while decreasing the false alarm rate which is related to the fact that the presence of noise, the contamination of the background statistics with the signature of the anomaly and the supposition of background homogeneity increase considerably the false alarm rate. The second challenge is related to the detection of anomaly with different shapes and sizes. In fact the size of anomalies can range from sub-pixel level to few pixels and the detection of these different sizes anomalies with the same detector steels a big challenge. The third challenge aims to achieve the nearest computational cost to real-time processing to perform anomaly detection on board as pixels are received.
Key Terms in this Chapter
Segmentation Based Anomaly Detectors: Anomaly detection approaches that are based on the segmentation of the hyperspectral image for the discrimination of the background pixels and then the deduction of anomalies.
Feature Selection Based Anomaly Detectors: Detectors that start by selecting features from the image, before anomaly detection, and use these features for the differentiation between anomalies and the background.
Kernel Based Anomaly Detectors: Detectors that use kernel strategies to separate the few pixels of the image that diverge from the background.
Statistic Based Anomaly Detectors: Anomaly detectors that use statistical metrics to calculate the difference between pixels of the background and anomalies.
Hyperspectral Images: A satellite image collected over hundreds of spectral bands that are narrow over a continuous spectral range. Every pixel illustrates the behavior of objects sun light reflection.
Anomaly detection: A blind detection of objects from hyperspectral images that deviate from the rest of the image.
RX Detector: An anomaly detection approach that took the name of its founders Reed and Xioli. It consists in calculating the statistical difference between the image pixels to look for anomalies.