Special issue-Computational optical imaging technology
2022, 51(2): 20220110.
doi: 10.3788/IRLA20220110
Computational optical imaging is an emerging research field to realize specific imaging functions and characteristics by jointly optimizing optical systems and signal processing. It is not a simple supplement to optical imaging and digital image processing, but rather an integrally combination of optical modulation at the front end (physical domain) and information processing at the back end (digital domain), where images and information are obtained through optical coding and mathematical modeling of the illumination and imaging system in a computationally reconfigurable manner. This new imaging mechanism is expected to break the limitations of traditional optical imaging technology on the optical system and image detector fabrication, manufacturing, operating conditions, power consumption, and cost, and significantly improve imaging function (phase, spectrum, polarization, light field, coherence, refractive index, 3D morphology, depth of field, blur recovery, digital refocusing, change of view angle), performance (spatial resolution, temporal resolution, spectral resolution, information dimension, sensitivity), reliability, and maintainability. At present, computational optical imaging has been developed into an emerging interdisciplinary research field that integrates geometric optics, information optics, computer vision, digital image processing, modern signal processing, etc., and has become an international research focus and hotspot in the field of optical imaging, representing the future development direction of advanced optical imaging technology. Many universities and research institutes at home and abroad are getting involved, making it a rapidly developing research field where "a hundred flowers bloom and a hundred schools of thought contend". As the first article in the column "Computational optical imaging technology" of the special issue "Nanjing University of Science and Technology" for the Journal Infrared and Laser Engineering, this paper provides a general overview of historical evolution and development status of computational optical imaging, and looks forward to its future development direction and the core enabling technologies on which it relies, to throw bricks and attract jade.
2022, 51(2): 20210891.
doi: 10.3788/IRLA20210891
In traditional pathology detection, the speed of diagnosis is limited due to the complex staining process and single observation form. The staining process is essentially associating color information with morphological features, and the effect is equivalent to that of biomedical images of modern digital technology. Sense segmentation, which allows researchers to greatly reduce the steps of biomedical imaging processing samples through computational post-processing, and achieve imaging results consistent with the gold standard of traditional medical staining. In recent years, the development of artificial intelligence deep learning has contributed to the effective combination of computer-aided analysis and clinical medicine, and artificial intelligence color transfer technology has gradually shown high development potential in biomedical imaging analysis. This paper will review the technical principles of deep learning color transfer, enumerate some applications of such technologies in the field of biomedical imaging, and look forward to the research status and possible development trends of artificial intelligence color transfer in the field of biomedical imaging.
2022, 51(2): 20220086.
doi: 10.3788/IRLA20220086
Microsphere-assisted super-resolution microscopy is an emerging technique which can be used to overcome the diffraction limit of conventional optical microscopes and significantly enhance their resolution. This technique is very promising for various applications because of the simplicity of its operation, its label-free and real-time imaging nature and its ability to be performed under white-light illumination with commercially available optical microscopes. Although there are many impressive results coming out along with the development of this technique, most studies are about the imaging properties, imaging quality improvement and manipulation of microspheres. A comprehensive theory on the super-resolution mechanism is still missing. Within this context, the progress of the microsphere’s imaging theory and the numerical methods in simulating the near-field focusing and far-field imaging phenomenon of microspheres was reported in this paper. The challenges and the future of this technique were also discussed.
2022, 51(2): 20220095.
doi: 10.3788/IRLA20220095
Computational optical microscopy imaging technology combines optical encoding and computational decoding to retrieve multi-dimensional information of microscopic objects through optical manipulation and image algorithm reconstruction, providing a powerful boost for microscopy imaging technology to break through traditional imaging capabilities. The development of this technology has benefited from the optimization of modern optical systems, image sensors, and high-performance data processing equipment, and is also enabled by the development of advanced communication technologies and equipment. As a highly integrated electronic device, the smartphone platform has an advanced image sensor and a high-performance processor, which can collect the image of the optical system and run the image processing algorithm, creating a new way for the realization of computational optical microscopy imaging technology. Furthermore, as a mobile communication terminal, the open operating system and various wireless network access methods of the smartphone platform endow the microscope with flexible and intelligent control capabilities and rich display and processing analysis functions, which can be used to realize diversified biological detection applications in various complex environments. In this paper, the computational optical microscopy imaging technology based on the smartphone platform was reviewed from four aspects. First, the design of the new microscopic imaging optical path based on the smartphone platform as an optical imaging device was reviewed. Next, the computational optical high-throughput microscopy imaging technology based on the advanced sensor of the smartphone platform was introduced. Then, the application of the data processing and interconnection capabilities of the smartphone platform in computational microscopy imaging was introduced, and finally some of the existing problems and solutions of this technology were discussed.
2022, 51(2): 20220006.
doi: 10.3788/IRLA20220006
Three-dimensional (3D) shape measurement based on fringe projection was widely used in industrial manufacturing, quality testing, biomedicine, aerospace and other fields. However, due to the short exposure time of raster images acquisition process, 3D reconstruction results were usually affected by serious image noise in the scene of high-speed measurement. In recent years, deep learning has been widely used in computer vision and other fields, and has achieved great success. Inspired by this, we proposed a learning based approach for noise reduction with raster images. Firstly, we constructed a convolutional neural network based on U-NET. Secondly, the neural network was constructed to learn the mapping relationship between the noisy fringe images and the corresponding high quality wrapped phase during the training process. With proper training, this network can accurately recovered phase information from noisy fringe images. Aiming at off-line 3D measurement in fast moving scene, experimental results show that the proposed method can recover high-precision phase information by using only one raster image, and the phase accuracy is better than the traditional three-step phase shift method. This method can provide a practical and reliable solution for improving the accuracy of 3D measurement in high-speed scene.
2022, 51(2): 20210889.
doi: 10.3788/IRLA20210889
Imaging through scattering media with high fidelity is still one of the main challenges in imaging analysis of deep biological tissues and distant astronomical observations. The computational imaging method based on deep learning has made significant progress in reconstruction quality and other aspects. However, when the scattering media in the actual system is unstable and the structure of objects is complex, and the obtained scattering dataset for training is limited, the pure data-driven method cannot realize efficient reconstruction. An efficient imaging method was proposed in reconstructing complex objects through unknown thin scattering media with different statistical properties, which was based on the effective combination of the speckle correlation theory and the powerful data mining and mapping capabilities. More information had been unearthed with the redundancy of the speckles and had been fully used with the neural network. This method obtained high-quality recovery of complex objects with complex scattering scenes and the training set is limited. This approach can promote the applications of physics-aware learning in practical scattering scenes.
2022, 51(2): 20210887.
doi: 10.3788/IRLA20210887
As a simple linear translation invariant operation, convolution has been widely used in various fields of image processing, and the convolutional neural network derived from it is brilliant in the field of artificial intelligence. In order to deal with the problem of limited computing power of AI reasoning chip in the post-Moore era, optical neural network came into being. As one of the important research hotspots, optical convolutional neural network plays an important role in promoting the development of optical neural network. An optical convolution system was designed, based on the uniform light path formed by micro lens array and lens, the image carried in the light place was convoluted in two-dimensions. The system can complete simple image smoothing and sharpening in the optical path. When the spatial light modulator is used to realize the convolution kernel and input surface, the system can realize three convolution forms of various step sizes, and can also realize multi-channel three-dimensional convolution through multiple projection or flattening, thus laying a foundation for the realization of optical convolution neural network for complex image processing tasks.
2022, 51(2): 20210895.
doi: 10.3788/IRLA20210895
High harmonic generation (HHG) with orbital angular momentum in the extreme ultraviolet could be produced by the interaction between vortex ultrafast infrared laser pulse and gas medium. In this paper, Laguerre-Gaussian (LG) beam with nonzero radial node was used as the driving laser. And through computing the single-atom response with the quantitative rescattering model, distributions of intensity and phase of HHG in the near and far fields were obtained by solving the three-dimensional Maxwell’s equation in the medium and the Huygens’ integral in the paraxial approximation, respectively. With the increase of the radial node in the driving laser, it is indicated that the distribution of HHG intensity shows the multiple-ring structure, the radial-node structure appears in the distribution of HHG phase, and the spatial region of intensity distribution is decreased in the near field, but increased in the far field. The phase-matching analysis showed that maps of spatial coherence length of short- and long-trajectory HHG are very sensitive to the mode of driving laser, qualitatively consistent with the maps of evolution of HHG field inside gas medium, which explained the features of vortex HHG under the LG beam with nonzero radial node.
