Download Full Text (1.5 MB)


The science of hyperspectral remote sensing is based on taking a fraction of the electromagnetic spectrum and breaking it into numerous bands for theoretical analysis and computations. The combination of all wavelengths in a given spatial area builds complete spectral signatures for each specific material in the scene. Based on the spectral signature obtained from hyperspectral imagery, one can detect and identify objects more precisely compared to using only three bands information provided by a RGB camera. Hyperspectral sensors can also assist in automatic target detection in noisy backgrounds since objects vary uniquely from the natural background in absorbing and reflecting radiation at different wavelengths. In many cases, the objects that the human eye fails to capture can be differentiated and identified based on the unique hyperspectral signature. Unfortunately, the spatial resolution for hyperspectral sensors is still extremely coarse compared to modern high definition camera. Thus, we present a visibility improvement technique that will increase the spatial resolution of the captured hyperspectral image and improve the image contrast. In the proposed algorithm, the image spatial resolution is increased by integrating intensity information from multiple related spectral bands. Leveraging our prior expertise with single image super-resolution on RGB imagery, we exploit the band information of the hyperspectral image and develop an adaptive contrast enhancement technique to construct a high spatial resolution image. Specifically, the enhancement algorithm selects the pixel-wise intensities to maximize the pixel’s neighborhood contrast. To verify the effectiveness of the proposed technique, we use the Resonon Pika II hyperspectral camera, which provides 240 spectral channels that ranges from 400-900nm with 2.1nm spectral resolution, to capture real-life images and test the visibility improvement methodology in a variety of environments such as low illumination or over-exposure regions. The proposed technique aids in real-world applications such as object detection, recognition, and tracking.

Publication Date


Project Designation

Graduate Research

Primary Advisor

Vijayan Asari

Primary Advisor's Department

Vision Lab


Stander Symposium poster


Arts and Humanities | Business | Education | Engineering | Life Sciences | Medicine and Health Sciences | Physical Sciences and Mathematics | Social and Behavioral Sciences