Research On Optical Earth Observation Image Atmospheric Physical Degradation And Compensation

Because of its wide coverage and wide spectral spectrum,remote sensing images play an important role in many fields?However,many of the image analysis and application suffer from the visibility degration,since most of them assume that remote sensing images are captured under good weather.As the electromagnetic radiation needs to pass through the atmopshere before they can be accepted by the sensor,the radiation,absorption and scattering of atmospheric particles will inevitably interfere with the final image,which will result in image quality degradation and limit the application of remote sensing system..The atmospheric compensation method based on digitial image processing has wide range of application prospect and value,because it is costly and easy to use.The essence of the method is mathematical modeling problem of the degenerate model.The existing atmospheric degradation model usually assumes that the atmosphere is uniform and ignores the influence of multiple scattering of atmosphere.Therefore,it is necessary to further expand and deepen the degenerate model and construct the effective atmospheric compensation technique.In this paper,the atmospheric radiation transfer equation and the model of atmospheric radiation degradation is introducted firstly.Then the remote sensing image degradation model with multiple scattering effects and atmospheric non-uniform distribution is proposed.According to the proposed model,the image atmospheric correction technology is studied and a complete image defog algorithm flow framework is constructed.Aiming at the requirements of different For different remote sensing application requirements and characteristics,a variety of atmospheric degradation compensation algorithm is proposed.Finally,the objective quantitative evaluation method of image atmospheric degradation is explored.The image degradation model proposed in this paper improves the existing degraded model by spatial convolution and the introduction of atmospheric non-uniform term.The constructed atmospheric degradation compensation framework use the input image to obtain model parameters.The multiband image defog algorithm based on Angstrom exponent is proposed to optimize the transmittance estimation of each band of the image and improve the processing efficiency of multi-spectral remote sensing data.Remote sensing image defog based on Mean-shift block deconvolution,which is used to reduce the algorithm workload by image segmentation and block-deconvolution.Based on CUDA remote sensing image defog algorithm GPU acceleration,whose computational speed is improved by using the hardware parallel computing ability and the potiential data parallelism of the algorithm itself.The atmospheric distribution estimation algorithm based on remote sensing image is proposed,and the atmospheric distribution in the image scene is extracted by the image data.The method of quantitative evaluation of atmospheric degradation of images based on neural network model has a high correlation with atmospheric degradation due to the nonlinearity and adaptability of the model.A series of experimental results show that the model proposed in this paper is more consistent with the atmospheric degradation of the actual image,so the atmospheric compensation method based on this model can obtain a more sharp and clear recovery image,especially for the remote sensing image with different degrees of atmospheric degradation.The optimization algorithm proposed for different application requirements can effectively impove the efficiency of the algorithm while maintaining the effect.The proposed atmospheric distribution extraction algorithm can effectively extract the atmospheric distribution of remote sesning image.Through the evaluation of different images,it is verified that the quantitative evaluation method of image atmospheric degradation is consistent with the degree of atmospheric degradation and the subjective evaluation.