Super-Resolution Image Reconstruction Applied to an Active Millimeter Wave Imaging System based on Compressive Sensing

Alkus U., Ermeydan E. S., ŞAHİN A. B., ÇANKAYA İ., ALTAN H.

Conference on Millimetre Wave and Terahertz Sensors and Technology X, Warszawa, Poland, 11 September 2017, vol.10439 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 10439
  • Doi Number: 10.1117/12.2278022
  • City: Warszawa
  • Country: Poland
  • Keywords: super-resolution, image reconstruction, millimeter wave imaging, compressive sensing, spatial light modulator, FOCAL-PLANE ARRAY, RESOLUTION, RADAR, INTERFEROMETER, MICROWAVE, DENSITY
  • Middle East Technical University Affiliated: Yes


The development of passive and active millimeter wave imaging systems is progressing rapidly fueled by the need for many applications in the area of security and defense. Imaging schemes that may either utilize array detectors or single detectors in scan architectures offer suffer from poor resolution due to the longer wavelengths used and the limits of the optical system in terms of lens and mirror dimensions. In order to overcome this limit, super-resolution techniques can be employed to enhance the resolution of the imaging system. Here, a form of this technique based on oversampling is applied to reconstruct the image of a target which is acquired using compressive sensing based on scanning the image plane using randomly patterned masks with fixed pixel sizes. The mm-wave stand-off imaging system uses a 93 GHz center frequency source and heterodyne sub-harmonic receiver place in a bi-static configuration to image a target in reflection mode. The image of the target is projected onto a mechanically scanned spatial light modulator (SLM), which is a patterned two-dimensional mask that is translated along one axis. In order to improve the resolution of the image, the masks are shifted by half the pixel size (2.5mm). To enhance the resolution of the image, the patterns are shifted by smaller steps, thereby each pixel is oversampled and the resulting new pattern and detected intensity is fed into the CS algorithm to reconstruct the image of the target. After the image reconstruction process, sharper edges are observed for a circular object of 12mm diameter compared to the image acquired by whole pixel step scanning.