Imaging through dynamic scattering media with stitched speckle patterns

Seeing through scattering media is highly desired in many occasions, such as remote sensing through clouds, fog and haze, underwater detection, anti-terrorism surveillance, biological tissue imaging and clinical diagnosis. The speckle patterns formed by signal lights passing through the scattering medium are difficult to distinguish. In fact, the information is only disturbed but not lost. However, multiple scattering experienced by photons when propagating in scattering media breaks the point-to-point correspondence required by traditional imaging methods, requiring some new imaging methods to overcome this problem.

If the incident light is tilted or shifted in a certain range relative to the scattering medium, the speckle pattern formed by the emergent light will be shifted but highly correlated. This phenomenon, known as the memory effect, has been well applied in scattering imaging, producing a new imaging modality, i.e., speckle autocorrelation imaging.

When a target within the memory effect range is illuminated by incoherent light, the speckle pattern detected by the camera is equivalent to the superposition of the speckle formed separately by each point on the target, and can be regarded as the convolution of the target with the point spread function (PSF). According to the convolution theorem, the Wiener-Khinchin theorem, and the fact that the autocorrelation of the PSF is a peaked function, the target image can be reconstructed from the speckle pattern by using an iterative phase retrieval algorithm. This method requires enough sampling information and a certain spectral resolution to ensure the reconstruction quality, so the previous experimental models have relatively small ratio of image distance to object distance. An intuitive solution to increase the image distance is enlarging the sensor area and/or employing a large aperture lens, which however is expensive and space-consuming.

The research group led by Dr. Honglin Liu from Shanghai Institute of Optics and Fine Mechanics proposed a method for the speckle sampling and processing, which has good adaptability and lower hardware dependence. The research results are published in Chinese Optics Letters, Vol. 18 Issue 4, 2020 (Xin Wang, Honglin Liu, Meijun Chen, et al. Imaging through dynamic scattering media with stitched speckle patterns[J]. Chinese Optics Letters, 2020, 18(4): 042604).

This method utilizes the equivalence between the spatial and temporal ensemble averaging of speckles in dynamic scattering system, acquiring many speckle patterns with different PSFs in sequence and stitching them together to get a big speckle pattern. The stitched pattern can satisfy the requirements of ensemble averaging and spectral resolution of discrete Fourier transform simultaneously.

While preserving the advantages of simple, fast and non-invasive of the speckle autocorrelation imaging, this method guarantees successful images reconstructions in larger image distances without any hardware upgrade. This method is helpful to expand the applications of speckle autocorrelation in remote sensing, underwater detection and so on. Future work will focus on further improving the detection distance and optimizing the reconstruction quality, as well as the reconstruction of multispectral targets.

Schematic diagram of imaging through dynamic scattering media with stitched speckle patterns