Moving Object Detection In Turbulence Degraded Image Sequence

Turbulence distortion seriously reduce the image quality,distortion caused by pixel distortion can bring about to the image of the target deformation or scaling,increasing the difficulty of the image sequence of the target detection and identification.There are different degrees of turbulence distortion in the remote imaging system.The difficulty of moving target detection in turbulence distortion image is that turbulence distortion is mixed with noise and other disturbances.The observed image is complex and it is difficult to correct the motion and turbulence of the target object.Resulting in the irregular movement of the entire image being distinguished.This paper focuses on the problem of target detection in the turbulence distortion scene.The main contents of this study include the following aspects:Firstly,the turbulence distortion image modeling and simulation are completed.The spatial transformation model is used to model the turbulence distortion,and the degree of distortion is described by quantifying the parameters in the spatial transformation model.Secondly,discuss the image turbulence distortion correction.The image is corrected by using the distortion correction and phased distortion correction scheme based on the distortion symmetry constraint.The feasibility of the correction algorithm is verified by experiments.The performance of the distortion correction algorithm is evaluated by the similarity measure between the distorted image and the corrected image.A new method of correcting the algorithm is proposed.A method of distortion correction using optical flow is proposed.The image sequence is corrected with the approximate reference frame obtained by simple averaging.After the iteration,the image sequence can be completed Correction.Then,target detection method based on the distortion correction is discussed to complete the target detection task under the turbulence distortion scene.The target detection algorithm based on distortion correction firstly removes the turbulence distortion in the image.The problem can be approximated to the problem of moving object detection in the image sequence which is not affected by turbulence distortion.For the corrected image sequence,the traditional target detection method is used to detect the object,it can be validated that the feasibility of the algorithm based on distortion correction based on the background elimination algorithm.Finally,the problem of image information loss caused by turbulence distortion correction is detected using the target detection method based on image content motion analysis.The composition of the observed image is divided into turbulence distortion components,relatively static background components and moving target components.According to the prior knowledge of each component,different constraints on the different components in the image are applied,and the distortion correction and The goal detection problem is unified as an optimization of the energy function,which avoids the loss of information caused by the distortion correction phase and completes the distortion correction and target detection task.