32
raster scanning of the object or the THz beam. In addition, one would like to preserve
the superior detection sensitivity of a single-point detector such as photoconductive
antennas (rather than the lower sensitivity provided by existing multi-pixel arrays)
and the simplicity and spatial coherence of a point-source transmitter. This chapter
describes a single-pixel THz imaging system based on an advanced signal processing
theory called compressed sensing (CS) [8,9], which enables both of these objectives.
In contrast to the THz Fourier imaging sestup using CS and phase retrieval described
in Chapter 3, this system does not require mechanical scanning of the THz receiver
on the image plane.
The rationale behind the improved acquisition speed of the single-pixel terahertz
camera is twofold. First, this camera performs compression simultaneously with im-
age sampling by modulating the spatial profile of the THz beam with a set of random
patterns, a technique enabled by CS. This imaging scheme requires significantly fewer
samples than the total number of image pixels to fully reconstruct an image, thus
speeding up the acquisition process [8,9]. Second, the speed of most existing THz
imaging systems is limited by the need to mechanically raster-scan the object (or the
THz beam) [1]. The method described in this chapter replaces this mechanical scan-
ning with the spatial modulation of the free-space THz beam, which can in principle
be much faster.
4.1 CS imaging principle
The principle behind the design of CS imaging systems can be summarized in the
equation y = Φx, where y is an M× 1 column vector of measurements, x is an image
with N2 pixels ordered in a N2 × 1 vector and the measurement matrix Φ is M × N2.
Using CS, we acquire a much smaller number of measurements than the number of