Research On Image Reconstruction Based On Laser Tomography Range Spectrum

Imaging through scattering media is a research focus in the field of target detection and imaging in recent years,which has an important application prospect in military,search and rescue,transportation and many other fields.In imaging processes,scattering media along the light path,such as cloud,fog and smoke,degrade the target image due to the effects of attenuation,scattering and backscattering in the active imaging.Range-gated imaging can remove additive backscattering noises by receiving signals of certain ranges;however,it fails to weaken the multiplicative interferences such as attenuation and scattering.In order to eliminate the multiplicative interferences in the process of imaging through scattering media,based on laser tomography range-spectrum imaging,methods and processes of target detection and image reconstruction are researched in this paper.The main works are as follows:1.The principles of range spectrum and the processes of laser longitudinal tomography are presented.Referring to light spectrum,the concept of range spectrum is introduced.In active imaging systems,the target/medium signal energy received by the detector varies with the target/medium range in the form of “spectrum”.Therefore,range spectrum is defined as the intensity-range distribution of echo signals;and the images captured by laser longitudinal tomography with increasing gate delays are defined as range-spectrum image sequences.The range-spectrum image sequences contain multiple information of targets and scattering media.The theoretical model of laser tomography range-spectrum imaging is built,and the energy equations of targets and medium layers are derived.The optical configurations and signal timings of the system are analyzed.2.Based on laser tomography range-spectrum imaging,a target detection technique together with an image reconstruction method is proposed.The range-spectrum image sequences record the echo signals of the entire light path.The proposed method extracts information of targets and scattering media from the image sequence to establish physical relationship between the scattering medium signal and the target signal,which can help to estimate and eliminate the degradation caused by scattering media.Then,the real target reflectivity can be acquired with help of this optical process.3.The degradation model of target images is put forward,and a variational method for the target image reconstruction is developed.According to the energy equations of targets and Monte Carlo simulation analyses,when the dual-pass transmittance is more than 0.01,target images are mainly degraded by the attenuation caused by medium layers,while the scattered photons have little effect on target images.The degradation model of target images is put forward,and the degradation caused by medium layers is described by a multiplicative operator,i.e.,degradation matrix.Then,the reconstruction process can be divided into two steps: first,estimating the degradation matrix;second,solving the inverse problem to obtain the retrieved target images.4.Estimation methods of degradation matrixes under various medium distributions are developed.For different medium distributions,we introduce different prior information to establish physical relationship between the medium layer signal and the target signal and to estimate the corresponding degradation matrix.When media surround targets,the estimation method based on medium layer close to target is proposed;When the medium layer is located between the target and the transceiver system,the estimation methods,based on target self-calibration and parameters pre-calibration,are respectively put forward to estimate the degradation matrix.5.The validity and feasibility of the proposed methods are verified by experiments.The range-spectrum image sequences are captured by laser longitudinal tomography system.The proposed methods can weaken the interferences of medium layers remarkably and retrieve target images.we compare the proposed method with the blind digital image restoration algorithms such as the homomorphic filtering.The proposed method can improve the PSNR(peak signal to noise ratio)and SSIM(structural similarity)values more effectively.Laser tomography range-spectrum imaging can receive echo signals from certain ranges so as to remove the background and backscattering noises.More important,this technique can utilize medium images in the range-spectrum image sequences of the entire light path to establish physical relationship between the medium layer signal and the target signal,and then estimate the degradation matrix and reconstruct the target image.Therefore,the proposed laser tomography range-spectrum imaging is essentially different from traditional range-gated imaging.According to the range-spectrum image sequences,multiple target/medium information can be extracted to establish the signal relationship.These information and relationship can be used for target image enhancement,as well as for noise suppression and three-dimensional target reconstruction.This technique can effectively enhance performances on target detection and imaging through scattering media.