Non-invasive scattering imaging method based on speckle estimation and deconvolution

Abstract

The invention discloses a non-invasive scattering imaging method based on speckle estimation and deconvolution. The non-invasive scattering imaging method comprises the steps that an integrated intensity matrix IIM<s> of N training samples O<s>(i=1,2,...,N) through the same scattering medium blocking area is acquired by using a non-invasive imaging system, and the reconstruction result O<-><s> of the training samples is obtained through processing; a constrained least square model is established by using a speckle pattern S and the convolution relationship between the training sample reconstruction result and the integrated intensity matrix obtained in the previous step so as to obtain the estimated speckle pattern S<^>; and the integrated intensity matrix IIM<c> of the sample O<c> to be observed in the scattering medium blocking area is acquired to perform deconvolution so as to reconstruct the image O<^><c> of the sample to be observed. The estimated speckle pattern can be obtained in a non-invasive mode so as to realize efficient reconstruction of complex samples and effectively enhance the resolution and the robustness of the reconstruction result.

Description

technical field [0001] The invention relates to the field of computer photography, in particular to a non-invasive scattering imaging method based on speckle estimation and deconvolution. Background technique [0002] Imaging through the scattering layer is a hot issue that needs to be solved urgently, and it has great application prospects in the field of medical diagnosis such as biological tissue imaging. However, when the scattering ability of the scattering layer is such that it can scatter almost all the light passing through it, it will become quite difficult to reconstruct the objects behind the scattering layer. [0003] For scattering imaging through strong scattering layers, the current imaging schemes mainly include the following four methods: wavefront trimming method, phase recovery method based on speckle scanning, phase recovery method based on single acquisition, and deconvolution based on PSF acquisition method. Among them, the wavefront trimming method u...

AI Technical Summary

Problems solved by technology

Among them, the wavefront trimming method uses a spatial light modulator to compensate the phase disturbance caused by the scattering layer to the outgoing light, so that a point on the object plane in the system will correspond to a single point on the phase plane, so that it can be directly imaged like a traditional optical imaging system; However, this method needs to pre-intrusively place a calibration point light source on the object surface, so as to use the adaptive optimization algorithm to obtain the compensation phase distribution of the spatial light modulator; therefore, this method cannot be non-invasive

The phase recovery method based on speckle scanning and the phase recovery method based on single acquisition can both non-invasively recover clear images of simple samples, but both methods rely on the performance of the phase recovery algorithm, and the phase recovery algorithm has different characteristics. Stable characteristics, in the case of non-invasive scattering imaging with large noise and error, the ability of phase recovery will decay rapidly with the increase of the complexity of the sample structure; in addition, the result of phase recovery will also lose samples The original position and direction information, so in these two algorithms, the autocorrelation of speckle is approximated as an ideal pulse function, which will cause the imaging accuracy to be limited by the size of speckle particles

The deconvolution method based on PSF acquisition can more accurately reconstruct the structural details of complex samples. However, this method also requires pre-invasive placement of a calibration point light source on the object surface to acquire the PSF of the system. Intrusive

Images