Fusion of Synthetic Aperture Radar Images and Optical Images Using Curvelet Transform and Retina Model

In recent years, various image integration techniques have been developed to improve their quality. In this paper, some image integration techniques such as Intensity-Hue-Saturation (HIS), Brovey transform, feedback, non-feedback retina model, wavelet transform, and curvelet transform are investigated to improve the spectral and spatial information of satellite images. Also, a new algorithm has been proposed to improve the image quality resulting from the combination of SAR and visiblelike images. In the proposed method, the curvelet transform is first applied to the three input levels of Synthetic Aperture Radar (SAR) and visible-like images, then using horizontal cells in the feedback retina model, spectral and spatial information below a specified and adjustable frequency is determined by a Gaussian low-pass filter and replaced with the curvelet coefficients of the integrated image approximation sub-band. Moreover, fine1 and detail1 sub-bands are selected from the visible-like image, and the coefficients of fine2, detail2 sub-bands are weighted and aggregated from both SAR and visible-like images in a specific way. Spectral and spatial quality evaluation criteria including Quality Index (Q_I), Measure the Quality of edges (Q^(AB/f)) Relative Dimensionless Global Error in System (ERGAS), Mutual Information (MI), Euclidian Distance (ED) and Standard Deviation (STD) were used to compare and analyze the results of the methods. The results of this evaluation indicated the remarkable performance of the proposed method in preserving the spectral and spatial information content of the integrated image compared to other methods.