Sub-pixel motion estimation for terahertz imaging

摘要

THz imaging is being increasingly applied to produce images of the interiors of objects and to identify variations. THz detectors are steadily improving, but lag behind visible light systems in terms of their resolution, i.e. due in part to the limited pixel number and the size of the array detector area. Synthetic aperture methods seek to increase the numerical aperture (and so image resolution). For example, the position of the detector is altered, and multiple images are captured in order to obtain the entirety of the image plane information. High precision registration and image fusion algorithms are then required to stitch together the individual images captured. Here, we propose a novel subpixel estimation method for THz imaging, which enables real-time operation with high temporal and spatial resolution. A THz imaging system is implemented using a continuous-wave THz source emitting at 300 GHz and a THz camera with 16×16 pixels. The subpixel estimation method is applied to the resulting THz images. We demonstrate that this method can be used to calculate and calibrate image position in a conventional THz imaging system, with significantly reduced computational expense.