Research On High-efficiency Near-field Terahertz Single Pixel Imaging Technology

Terahertz(Terahertz,THz)wave refers to electromagnetic radiation with a frequency of 0.1-10 THz(corresponding to a wavelength of 3 mm-30 ?m).It is in the special position of the transition from electronics to photonics in the electromagnetic spectrum and has the characteristics of perspective,safety,spectral resolution,and so on.Terahertz imaging is widely used in security inspection,industrial inspection,and biomedical fields due to its characteristics.In the traditional terahertz imaging system,the imaging is performed by scanning the sample point by point,which causes problems such as long imaging time and unstable system.At present,although terahertz area array detectors have made great progress,commercial terahertz cameras are still relatively expensive for most researchers.Therefore,exploring advanced terahertz imaging technology is the research focus of researchers.In recent years,single-pixel computing imaging has developed rapidly and has become one of the important research hotspots in the field of optics.In single-pixel computing imaging,the intensity of the spatially-encoded target image is collected by a single-point detector,and the image information is calculated back using a known encoding pattern.The key components of single-pixel computational imaging are spatial light modulators and reconstruction algorithms.Unfortunately,commercial spatial light modulators work in the visible and infrared bands,and cannot be directly applied in the terahertz band.Although light-controlled semiconductor technology can modulate terahertz waves by changing the spatial distribution of carriers,the efficiency of terahertz light-controlled modulation based on traditional intrinsic semiconductors is generally not high,which directly affects the imaging quality and studies efficient terahertz light Controlled modulator is an important link to improve single-pixel terahertz imaging quality.In terahertz single-pixel computational imaging,compressed sensing algorithms are widely used for their reliability.Although the algorithm can reconstruct the image signal only by partially acquiring the coding intensity,the image quality deteriorates as the sampling rate decreases.Therefore,it is important to study high-quality image reconstruction algorithms at high sampling ratios.In addition,terahertz imaging suffers from poor image quality due to the influence of long-wave diffraction.Although terahertz single-pixel calculation super-resolution imaging solves the problem of poor resolution,high-energy laser illumination is often used in order to balance modulation efficiency.This paper focuses on the problems of terahertz imaging quality and time and proposes a new imaging method that takes into account both cost and performance.Firstly,an efficient terahertz optical control modulator was developed to enhance the modulation efficiency and reduce the system complexity,and a terahertz single-pixel computing imaging system was built.Secondly,an efficient basic scanning imaging algorithm is studied to reduce the time cost,and finally,a single-pixel terahertz nearfield imaging is realized by combining a high-efficiency modulator and an imaging algorithm.The main contents and innovations of the paper are as follows:(1)Preparation and characterization of high-efficiency optically controlled terahertz modulators.The modulation efficiency of terahertz is directly related to the carrier concentration.Free semiconductor photocarriers will disappear due to recombination.In order to obtain efficient terahertz modulation,this study proposes two methods to increase the intrinsic semiconductor carrier concentration: heterojunction and physical passivation.Method one: Use the potential barrier difference between graphene and silicon to separate the electrons and holes of photoinduced free carriers,reduce the recombination probability,and increase the carrier concentration.The experimental results show that under the same illumination,the terahertz modulation factor of silicon-based graphene is 3 times that of the traditional high-resistance silicon modulation factor.Method two: prepare the silicon dioxide passivation layer by high-temperature thermal oxidation process to reduce the surface recombination center and at the same time study the efficiency of the silicon dioxide layer with different thicknesses to the light control terahertz modulation.The experimental results show that when the 808 nm continuous laser is incident at 25 °,the thickness of the silicon dioxide is 148 nm,and the maximum value of the terahertz modulation occurs.Compared with the traditional high-resistance silicon wafer,the modulation factor increases by about 2 times and the performance is stable.(2)The construction of a terahertz single-pixel computing imaging system.Due to the use of two-dimensional raster scanning in traditional terahertz imaging,its imaging speed is severely limited by the image size.Using a compressed sensing imaging algorithm,combined with a terahertz time-domain spectroscopy system,spatial undersampling terahertz spectral imaging is realized.Combining the compressed sensing algorithm with the scanning motor,the designed acquisition strategy can effectively reduce the number of samples and realize the spectral reconstruction of the image.Experimental results show that when the sampling rate is 50%,terahertz images can be reconstructed,which saves half of the sampling time and storage space.In addition,combining the terahertz time-domain spectroscopy system and the lightcontrolled modulator,the effects of optimized silicon wafers and traditional silicon wafers on terahertz imaging are compared.The experimental results show that no matter whether it is silicon-based graphene or passivated silicon wafer,continuous terrestrial laser irradiation can reconstruct a clear terahertz image,while the image reconstructed by high-resistance silicon as a modulator is blurred.An imaging system that combines a compressed sensing imaging algorithm and an efficient terahertz modulator can save 75% of the sampling time.(3)Research on efficient single-pixel imaging algorithm.Using the terahertz single-pixel imaging system of the light-controlled modulator,efficient single-pixel imaging algorithms are studied: Fourier single-pixel imaging and Hadamard singlepixel imaging.The projected orthogonal mask is used to obtain the coefficients of the image in the transform domain,and the inverse transform is used to obtain the image reconstruction.Unlike random sampling in compressed sensing,since the natural image is sparse in the transform domain and the energy is concentrated in the low-frequency region,high-quality image reconstruction can be achieved by collecting the modulation signal of low-frequency fringes.Experimental results show that only about of 10% sampling time is required to reconstruct high-quality images.(4)Realization of terahertz single-pixel near-field imaging.Terahertz imaging is affected by long-wave diffraction,and its image resolution is difficult to meet actual needs.By making a micrometer-thickness light-controlled modulator,the imaging target is placed on the back of the modulator to achieve near-field imaging conditions and combined with Fourier single-pixel imaging algorithm,fast and high-quality image reconstruction is achieved.Experimental results show that using a sampling time of about 10% can reconstruct a terahertz image with a resolution of ? / 5.