Optics Communications 413 (2018) 269–275
Contents lists available at ScienceDirect
Optics Communications
journal homepage: www.elsevier.com/locate/optcom
Single-pixel non-imaging object recognition by means of Fourier spectrum
acquisition
Huichao Chen, Jianhong Shi *, Xialin Liu, Zhouzhou Niu, Guihua Zeng
State Key Laboratory of Advanced Optical Communication Systems and Networks, Shanghai Key Laboratory on Navigation and Location-based Service, and Center of
Quantum Information Sensing and Processing, Shanghai Jiao Tong University, Shanghai 200240, China
a r t i c l e i n f o
Keywords:
Data processing by optical means
Computational ghost imaging
Image recognition
Optical processing
a b s t r a c t
Single-pixel imaging has emerged over recent years as a novel imaging technique, which has significant
application prospects. In this paper, we propose and experimentally demonstrate a scheme that can achieve
single-pixel non-imaging object recognition by acquiring the Fourier spectrum. In an experiment, a four-step
phase-shifting sinusoid illumination light is used to irradiate the object image, the value of the light intensity is
measured with a single-pixel detection unit, and the Fourier coefficients of the object image are obtained by a
differential measurement. The Fourier coefficients are first cast into binary numbers to obtain the hash value. We
propose a new method of perceptual hashing algorithm, which is combined with a discrete Fourier transform to
calculate the hash value. The hash distance is obtained by calculating the difference of the hash value between
the object image and the contrast images. By setting an appropriate threshold, the object image can be quickly
and accurately recognized. The proposed scheme realizes single-pixel non-imaging perceptual hashing object
recognition by using fewer measurements. Our result might open a new path for realizing object recognition
with non-imaging.
© 2017 Elsevier B.V. All rights reserved.
1. Introduction
Single-pixel imaging techniques are often referred to as ghost imag-
ing (GI). As a new technology and an intriguing method, over the past
20 years, ghost imaging has attracted great attention and achieved
significant development. In the 1980s, the former Soviet Union scholar
Klyshko proposed a ghost imaging scheme according to the entan-
glement behavior of spontaneous parametric down-conversion photon
pairs [1]. A research team at the University of Maryland realized ghost
imaging based on an entanglement source in 1995 [2,3]. Later, scholars
confirmed that pseudothermal light and thermal light can also be used in
ghost imaging [4–12]. In 2009, Bromberg realized computational ghost
imaging [13–15] through a spatial light modulator (SLM) preset light
source. In recent years, some new ghost imaging schemes have been
proposed. These include differential ghost imaging (DGI) [6], compres-
sive sensing ghost imaging (CSGI) [5], correspondence ghost imaging
(CGI) [16,17], sinusoidal ghost imaging (SGI) [18], and Fourier ghost
imaging [19]. Ghost imaging can break through the diffraction limit to
achieve high-resolution imaging [20,21]. In view of the above research,
ghost imaging has potential applications in remote sensing [22,23],
*
Corresponding author.
E-mail addresses: huichao.chen@sjtu.edu.cn (H. Chen), purewater@sjtu.edu.cn (J. Shi).
image encryption, weak light detection, and the imaging of penetrating
scattering media.
In recent years, the focus of research on ghost imaging has in-
creasingly turned from basic research to practical applications [24],
especially with regard to interdisciplinary applications. Previous work
realized object authentication in other ways through ghost imaging
[25–28]. The proposed methods promoted the further application of
ghost imaging in the direction of object recognition. It is necessary to
reconstruct the object image in order to realize object recognition. The
imaging process not only increases the complexity of the computation
but also lengthens the time required for object authentication. However,
to our knowledge, no single-pixel technique has successfully achieved
non-imaging object recognition thus far.
The perceptual hashing algorithm (PHA) [29–33] is a hash algorithm
that is mainly applied in the search for similar images. Perception
hashing technology converts image data into thousands of binary se-
quences [29]. It is a promising and effective method to solve image
content authentication. Specifically, PHA generates a ‘‘fingerprint’’ for
each image. Traditionally, a metric must be defined to measure the
distance between ‘‘fingerprints.’’ The hash distance metric used in the
https://doi.org/10.1016/j.optcom.2017.12.047
Received 23 October 2017; Received in revised form 13 December 2017; Accepted 17 December 2017
Availableonline 4 January 2018
0030-4018/© 2017 Elsevier B.V. All rights reserved.