Research On Optical Image Edge Detection

Imaging is one of the most important application fields in optics.The exploration of optical imaging by human beings can be traced back to the optical phenomenon of "pinhole imaging" recorded in the Book of Mo Jing during the Warring States Period of China.Later,with lens structures proposed and improved,it further promoted the development of geometric optics,enhanced the control of light wavefront,and also opened up a variety of new imaging methods and fields.With the development of nanomachining technology,optical metasurfaces composed of sub-wavelength nanostructures have a broad application prospect in imaging technologies,such as color holography,broadband achromatic imaging,and super-resolution technology,due to their unique advantages of lightweight,planarization and easy integration.Images are the basis of human vision.Compared with flat areas of images,human eyes are more sensitive to edge information.Therefore,image edge information plays a significantly important part in image analysis and human vision.Edge detection,as an effective method to represent the boundary of objects,is a basic tool in image processing,machine vision and computer vision.In recent decades,great progress has been made in optical information processing technology,which also promotes the wide application of image edge detection technology in face recognition,high contrast imaging,astronomical observation,autonomous driving and other fields.Compared with digital image processing,image spatial differentiation based on optical analog computing has the unique advantages of fast-speed operation and low power consumption,which could be used for real-time,parallel,high-speed and large-capacity image processing.This thesis mainly introduces optical image edge detection technologies based on filters including metasurfaces and spiral phase plates.We carried out some related work of quantum image edge detection and nonlinear imaging with field-of-view enhancement and edge enhancement,containing both isotropic two-dimensional edge detection and anisotropic one-dimensional edge detection.In this thesis,spiral phase contrast imaging and edge detection based on the Spin-Hall effect are introduced,as well as the brief theoretical analysis,simulation and experimental exploration.The research content of this thesis is primarily divided into the following parts:1.The spiral phase contrast imaging based on the nonlinear sum-frequency process is experimentally studied,and the visualization of the edges of objects under infrared illumination is realized.We use a quasi-phase-matching crystal to focus the infrared signal light(1550nm)carrying the object information into the crystal.The optical vortex(792nm)of topological charge l=1 is used to pump the crystal,and the spectral information of the object is modified in Fourier space.Finally,the infrared image invisible to human eyes is converted to the visible(525nm),and the object contour is lit at the same time.2.The dynamic evolution process of image field-of-view using phase-matching conditions in nonlinear imaging is studied.By accurately adjusting the temperature of the crystal,we enhanced the conversion angle bandwidth of the signal light,and finally the up-converted image achieved the maximum adjustment range of the field of view(FOV)of 2.1 times in both the conventional mode and the edge enhancement mode.Based on the split-step Fourier method and Collins diffraction integral formula,we have done a series of theoretical calculation and numerical simulation,and the constructed model is in good agreement with the experimental results.3.Real-time(0.5Hz)quantum edge detection of objects under ultra-low light level illumination is performed by combining heralded single photon imaging and spiral phase contrast imaging.We use a 405nm narrowband laser to pump a 2cm-long PPKTP crystal,generating time-energy entangled photon pairs with a spontaneous parametric down-conversion process.A single photon avalanche diode detector(SPAD)is used to detect photons at the heralding arm;an intensified charge coupled device(ICCD)is used to detect photons at the imaging arm.By switching vortex filters with different topological charges,we realize global edge detection and curved edge detection of objects respectively.and verify the time correlation characteristics of the quantum image and the shadow effect caused by the transverse displacement of the filter.Moreover,compared with direct single photon imaging,the signal-to-noise ratio(SNR)of the image using heralded single photon imaging can be greatly improved under a low light level.4.A quantum optical image edge detection switch,which can remotely switch image state,is demonstrated based on the high-efficiency dielectric metasurface and high-quality polarized entangled light source.In the experiment,we use the laser written dielectric metasurface as an optical image differentiator,and combine lenses and polarization elements to form an edge detection system.At the same time,we have prepared a high-fidelity polarization-entangled light source,which can selectively trigger the edge mode or normal mode of the image acquired by ICCD at the imaging arm by projection measurement of the polarization state of heralding photons.The switching method of measurements doesn't require any operation on the imaging system.Due to the strong second-order time correlation characteristics of entangled photon pairs,the images acquired by the coincidence measurement method have a higher signal-to-noise ratio compared with its counterpart using the traditional internal trigger direct detection method.The main highlights and novelty of this thesis are as follows:1.The traditional spiral phase contrast technique is extended to the nonlinear frequency conversion fields,and the edge visualization of the intensity object under invisible light illumination is subtly realized experimentally,which can significantly "lighten" the outline of the object illuminated by infrared light.We use the method of frequency conversion to convert the infrared image information to the visible spectrum,and then use the visible image detector with strong performance and low price for edge image acquisition,which overcomes the shortcomings of the infrared image detector with poor sensitivity,low efficiency and high price.2.One method is proposed to accurately control the image FOV by controlling the crystal temperature.In the nonlinear imaging process,we achieve the maximum imaging field-of-view adjustment range of 2.1 times in both the edge image and normal image.Compared with other nonlinear field-of-view control methods,such as a broad-spectrum light source,dual-wavelength illumination,designing temperature gradient of the crystal,rotating crystal angle,etc.,the operation is simpler and more convenient,and the requirements for light sources are lower.This new technique demonstrates for the first time the parallel image processing method of edge enhancement and FOV enhancement.And it has important potential applications in biological imaging,pattern recognition and infrared remote sensing.3.An efficient real-time(0.5Hz)edge detection method based on a quasi-phase-matching PPKTP crystal is proposed.In addition to global isotropic edge enhancement,we also verify the extraction and recognition of specific features of objects,including right-angle edge enhancement and orientation-selective shadow effect.Using the compact time correlation characteristics of photon pairs and the photon counting mode of the ICCD camera,the experimental results show the advantages of high fidelity and low noise without post-processing.This work provides a new solution for high contrast imaging of living biological samples at a low light level.4.The polarization-entangled photon source is introduced into the image edge detection system based on the dielectric metasurface to realize the remote nonlocal switching between normal mode and edge detection mode of the detected image state.And the two-dimensional nanostructures can be impressed onto optical glass using laser direct writing technology,which can be integrated into conventional optical components.This work is the first attempt at quantum metasurface research in controllable image edge detection.Since the image state is controlled by the post-polarization selection,it has potential applications in image encryption and steganography.Besides,the higher SNR will show certain advantages in the scene of photon-limited illumination,such as enzyme reaction tracking and the observation of living cells.This work will promote more related researches on the combination of quantum optics and metasurface materials.