Ultrasensitive and Rapid Multi-Dimensional Spectral Sensing at Standoff Distances by Non-Uniform Sampling and Statistical Reconstruction

Abstract - Approved for Public Release Research Problem: Rapid identification and analysis of complex chemical and biological targets represents an important aspect of US defense capabilities. However, remote sensing is often hampered by detrimental effects of obscurants such as cloud and fog that may dramatically reduce image quality in standoff detection scenarios. In addition, image contrast using conventional linear optical signatures is oftentimes poor owing to spectral overlap. Nonlinear optical methods can overcome both these challenges to obtain high-contrast images even in challenging environments. However, the disadvantages for this extreme specificity, is poor sensitivity, low dynamic range, and unreasonably long acquisition times, making such methods impractical for hyper-spectral imaging, especially at high frame rates. Objective: The objective of this program is to develop chemically sensitive standoff detection and imaging using nonlinear optical methods. The central premise underlying this approach is that quantum signatures encoded in the nonlinear optical response of the target leads to enhanced contrast, which may be used to differentiate a specific target within a large background. Technical Approach: We propose to employ non-uniform sampling and statistical reconstruction methods adapted for optical measurements, which can reduce acquisition times by 3-4 orders of magnitude, while reaching a dynamic range of over 104. A full two-dimensional (2D) spectrum may be recorded in less than 1 ms per pixel, which, when combined with sub-sampling imaging methods, may allow high-resolution imaging at video frame rates. Once a region of interest is identified, higher-order experiments may be required for improved chemical discrimination. For instance, a full 3D spectrum may be recorded in less than 100 ms, while the use of specialized hardware could reduce acquisition times even further to <10 ms per pixel depending on the particular target. One major advantage of nonlinear hyper-spectral imaging is the ability to ‘beat’ the diffraction limit of linear absorption or emission. Recently, we have shown that we can perform THz-Raman imaging within a microscope platform with a spatial resolution determined by visible, rather than THz light. This represents a x500 fold improvement in spatial resolution compared to direct THz absorption and emission, and with significantly higher sensitivity. Anticipated Outcomes: Specifically, we aim to develop methods to acquire spectral signatures in the far-IR to visible regions of the spectrum using a single, broadband visible/NIR light source. Critically, the high-frequency source will enable diffraction-limited imaging at sub-micron resolution with high-NA optics, or hundreds of micron resolution at standoff distances with low-NA optics, significantly better than the resolution achievable with direct low-frequency light sources. With the increased chemical specificity, however, comes decreased sensitivity. Therefore, we outline reconstruction methods to achieve ppm level sensitivity of the nonlinear signal relative to the linear response. The primary objective is to combine these methods to simultaneously achieve high-specificity, high-sensitivity, and high-speed for next-generation active standoff detection in challenging visual environments. Impact on DoD Capabilities: The proposed project is directly responsive to the ONR EO/IR Sensors and Sensor Processing Program whose objective is “to detect, classify/identify, and localize/geolocate air, sea-surface, and ground targets.”