ABSTRACT
This chapter provides a background to those unfamiliar with hyperspectral images and details the application of low-rank and sparse decompositions in hyperspectral video analysis. Traditional RGB images capture light in the red, green, and blue portions of the visible light spectrum. This is due to the way images are displayed on computer monitors and televisions using red, green, and blue pixels. Each layer represents the amount of radiated energy being emitted at a particular wave length. Images using more than 3 layers are referred to as multispectral or hyperspectral images. These images can involve light that is outside the visible spectrum, such as infra-red (IR) and UV (ultra-violet) light. Hyperspectral images have a higher spectral resolution compared to multispectral images while being limited to a narrow spectral bandwidth. By imaging the light that is absorbed and reflected in high detail within a certain region of the electromagnetic spectrum, it is possible to identify particular materials present in the image. One major application of hyperspectral imaging is the detection of invisible gaseous chemical agents and other anomalous harmful particles. This problem comes up in many practical applications such as defense, security, and environmental safety [1].
