Research On Quality Assessment Of Image Compression And Image Fusion For Optical Remote Sensing Image

Due to remote sensing satellite imaging system is imperfect in performance, image compression algorithm and remote data transmission equipment as well as the atmosphere and equipment noise interference, etc., remote sensing image degradation and fuzzy distortion is inevitable in the process of the remote sensing satellite image acquisition, compression and transmission. These bring a great deal of difficulty to the remote sensing image processing, analysis and application.This paper conducted in-depth research on the key technology of remote sensing image subjective quality evaluation and objective quality assessment against the practical problems of image compression subjective evaluation and objective evaluation in remote sensing image compression algorithm comparison Priority Selection and the problem of only the subjective qualitative description without objective and quantitative indicators in the current national standard remote sensing image quality evaluation. The paper explores quality evaluation methods and ideas of the original remote sensing images and remote sensing fusion image on the basis of the latest developments in the status quo of image quality assessment theory and techniques, and laid the foundation for further research. The main work is as follows:First, research on the image quality subjective evaluation methods.Against the practical problems of optimal selection on the remote sensing image compression algorithm, this paper proposed a new image quality subjective evaluation method based on Hill sorting algorithm, and developed a new remote sensing image subjective evaluation software based on the method. This method combines Hill sorting algorithm with the existing paired comparisons of remote image compressed subjective evaluation method, which had the advantages of both sorting methods and paired comparison method, and reduced the number of images comparing in the subjective evaluation. Subjective evaluation experiments through two professional frontline workers for the924remote sensing compression image were analyzed by using this software. The results show that the new method of subjective evaluation improves efficiency up to38%-52%. than the original paired comparison method. Second, Objective image quality assessment. In order to evaluate the quality of remote sensing compression distortion of the image from an objective point, this paper presents a new image quality evaluation model-gradient similarity (GSIM) model combined gradient amplitude, phase with structural similarity of the image (SSIM), and a new image quality evaluation algorithm based on the model. The new model, compared with SSIM model and gradient model, includes not only the brightness, contrast, and structure of the three parts information, but more importantly, this model increases the gradient phase information. Through experiments on982distorted image and924remote sensing image compression of the LIVE database, the results show the performance of the new model is better than MSE, PSNR, SSIM and other traditional models and gradient-based model. Compared with the SSIM and other models, the new model not only can solve the problem of serious distortion of the objective evaluation of the image not entirely consistent with the subjective perception, but also can better handle of the problem of the poor effect of mixed evaluation on various types of the distorted image.Third, conformational quality evaluation of remote sensing images productIn the plan view of the remote sensing images produced specification GB (GBT15968-2008), there is only subjective qualitative description ("rich layers, legible and uniform color, contrast moderate") with no objective and quantitative indicators. So the article breaks down the quality of remote sensing image into several specific areas such as clarity, the amount of information, the accuracy of radiation, texture information, spectral fidelity, spatial detail, SNR, etc., and gives a specific objective evaluation indicators to all aspects of the above on the basis of deep study on the existing objective indicators. Furthermore, aimed at subjective and qualitative description of moderate contrast, it also proposed a new contrast evaluation indicator based on the Gaussian-weighted on the basis of in-depth study of the characteristics of the human visual system. In the end, the experiment will verify the effectiveness of the new index.Four, evaluation of remote sensing image fusion methodFor the actual selection of remote sensing image fusion algorithm to remote sensing image processing production units, the paper deeply conducted research and comparison to the existing remote sensing image fusion algorithm. It studied the basic principles and algorithms of various fusion methods, and analyzed the characteristics of various fusion algorithm. Then, it gave selected principle of fusion algorithm based on the application of remote sensing images.