| In recent decades,with the rapid development of space remote sensing technology,remote sensing image has been widely used in commercial and military fields,and the high spatial resolution remote sensing image has valuable application value and research significance.Due to the influence of uncontrollable factors such as atmospheric disturbance and poor illumination conditions in the imaging process of remote sensing image,the quality of remote sensing image declines,which makes it very difficult to the follow-up work of quantitative analysis and mapping.In order to improve the quality of remote sensing images,the factors causing the degradation of remote sensing images are analyzed and summarized,and the image enhancement research is carried out for remote sensing images interfered by thin clouds and mist and low illumination remote sensing images under insufficient illumination.The main work contents of this paper are divided into the following aspects:Aiming at the low accuracy of remote sensing interpretation caused by contrast reduction,sharpness reduction and color imbalance taken in foggy weather,a remote sensing image enhancement method infused with adaptive contrast histogram equalization and multi-scale retinex with color restoration is proposed.CLAHE algorithm obtained more image information and suppressed the noise in homogeneous areas.MSRCR algorithm enhanced the details of dark areas and compensated the color distortion caused by contrast enhancement,which can achieve the effect of color restoration.The images are separately enhanced by CLAHE and MSRCR method,then the larger absolute value of the high-frequency components at the corresponding positions of the two images was taken as high-frequency coefficients,and the fusion method based on local variance-weighted was used to fuse the low frequency coefficient;finally,the fusion image was obtained by wavelet reconstruction.The experimental results show that the subjective visual effect and image quality index of fusion images are improved to a certain extent.According to the similarity between the inversion image of low illumination image and fog image,a dark channel prior of low illumination remote sensing image enhancement algorithm based on physical model was proposed.Different from the traditional low illumination image enhancement algorithms,that only carries out the simple transformation in the spatial domain and frequency domain,but enhances the image essentially based on the image degradation physical model.Aiming at the problems that the low efficiency of dark primary color algorithm and the lost edge information of enhanced remote sensing image due to "halo" effect,the algorithm combines the characteristics of remote sensing image and use guided filter instead of "soft matting" method to estimate the transmission rate so the processing speed was improved,and use the ideal high-pass filter to compensate the edge of remote sensing image,the enhanced remote sensing image has subjective visual effect and the objective evaluation was improved.The image enhancement algorithm in this paper is applied to engineering practice,and the fog removal algorithm of remote sensing image and the enhancement algorithm low illumination remote sensing image in this paper are used to remove fog and enhance the aerial photos of Songming County,Kunming City,Yunnan Province,to get the clear image data of Songming County,for land cover research.Based on the clear images,the enhanced remote sensing interpretation marker database was established,and the land cover in Songming County was divided into six categories,namely,cultivated land,grassland,woodland,construction land and water,by using the supervised classification method.The enhanced images were classified by different algorithms and were analyzed subjective method and objective method.The classification accuracy of the images processed by the improved algorithm was improved,which verified the practicability of the image enhancement method in remote sensing interpretation. |
PDF Full Text Request