Significance The development and application of many optical imaging and measurement systems have been promoted based on the principle of straight-line propagation of light, and the artificial intelligence technologies have also been developed rapidly. However, when the target to be detected is blocked by the scatterers such as clouds, haze, suspended dust, and turbid water with large optical thickness, due to the lack of point-to-point direct mapping between the object domain and the image space pixels, how to obtain the target image effectively becomes a difficult problem to be solved in the field of optical imaging.
The influence of scatterers on optical signal transmission mainly includes absorption and scattering, especially scattering. Scattering is due to a large number of scattering media or scattering particles with different refractive indexes and particle size distributions in the scattering body, which makes the light wave front from the target subject randomly interfere in the process of penetrating scatterers, resulting in the reduction of signal-to-noise ratio of the target and distortion of direct detection images. Optical imaging of penetrating scatterers is suitable for complex and diverse application scenarios, long target distance, and large optical thickness of scatterers. It is of great significance to solve the problem of optical scattering imaging with wide field of view and long distance.
Progress Various imaging methods based on ballistic light (non-scattering light) and non-ballistic light (scattered light) have been developed to solve the problem of optical scattering imaging. Scattering imaging technologies based on ballistic light acquisition, such as range gated imaging, polarization imaging, and adaptive optics imaging, have played important roles in astronomical imaging, transportation, underwater exploration, and biological imaging. For example, in 2017, Li et al. from National University of Defense Technology (NUTD) proposed a method for degraded matrix estimation and target image reconstruction based on laser longitudinal tomography, which effectively solved the multiplicative interference problems such as the non-uniform attenuation of the scattering medium to the target signal in range gated imaging. In 2019, Zhao et al. from Zhejiang University proposed a multi-guide-star conjugate adaptive optics correction method. By using multiple navigation satellites, the correction area of pupil adaptive optics method was increased, the correction efficiency was improved, and the field of view was expanded.
With the rapid development of computational imaging technology, a variety of new scattering imaging technologies have been developed by combining the scattering imaging with the computational imaging. In 2010, Popoff et al. proposed the scattering imaging technology based on optical transmission matrix, and in 2012, Bertolotti et al. developed the scattering imaging technology based on the optical memory effect (OME) and speckle correlation. In 2014, Katz et al. proposed the noninvasive single frame scattering imaging technology to overcome the mechanical instability of angle scanning and acquisition time-consuming defects, which promoted the development of computational scattering imaging technology.
Coherent diffraction imaging (CDI), ptychographic iterative engine (PIE), correlation imaging, non-line-of-sight (NLoS) imaging, and other new computational imaging methods are gradually combined with the requirements of scattering imaging, and have been developed rapidly, gradually solving many problems in the application of scattering imaging. In 2016, Zhou Jianying’s team from Sun Yat-sen University modeled the scattering imaging process as a convolution operation, realizing the field of view beyond the limited scope of OME. In 2018, Sahoo et al. from Nanyang Technological University in Singapore, based on the single frame speckle correlation imaging technology, used the speckle image generated in the area beyond the limited range of OME to realize scattering imaging of hidden objects in wide field of view. In 2019, Shao Xiaopeng’s group from Xidian University realized scattering imaging beyond the limited range of OME through the use of prior information; Dai Qionghai’s research group from Tsinghua University proposed and implemented a wide field of view speckle correlation imaging technology. In 2019, Li Xiujian’s group from NUDT proposed a single frame coherent power spectrum scattering imaging technology and a variable aperture Fourier PIE imaging technology, which effectively improved the imaging efficiency and enlarged the width of field of view; Shanghai Institute of Optics and Fine Mechanics proposed the idea of realizing scattering imaging in different regions to broaden the imaging field of view; Yao Baoli’s research group of Xi'an Institute of Optics and Precision Mechanics proposed a scattering imaging technology based on PIE and shower curtain effect, which broadened the imaging field of view. In 2018, Chen Pingxing’s research group from NUDT realized scattering imaging beyond the limited range of OME based on the optical path of correlation imaging. In 2019, Charles Saunders from Boston University in the United States proposed the computational perimetry using ordinary digital cameras. In 2020, Metzler at al. from Stanford University in the United States completed the NLoS imaging with remarkable effect by using the speckle correlation technology and deep learning method. These are beneficial attempts of computational scattering imaging.
Conclusions and Prospect In recent years, various optical scattering imaging technologies have been developed rapidly, have played important roles in biomedical microscopic imaging, military and civil target detection, aviation and road traffic monitoring, and other fields, and will play a greater role. However, there are still many problems in the application of wide field of view and long-distance scattering imaging. Combined with the use of ballistic and non-ballistic light, the computational scattering imaging technologies based on OME and speckle correlation, combined with CDI, PIE, and other computational imaging technologies, are expected to promote the development and application of wide field of view and long-distance scattering imaging. According to the technical development and application requirements, it is worth developing scattering imaging technologies which have the characters of the combination of active and passive imaging methods, facing non-sparse targets, moving targets, and dynamic scattering media, have three-dimensional imaging ability, or have color and spectral imaging capabilities. And it is necessary to develop the abstract mathematical model which can express the whole optical effect of scattering media.
