Research On Remote Sensing Image Fusion Algorithm Based On The Second Generation Crvelet Transformation

Curvelet is more suitable for a new multi-scale transform analysis of image processing than wavelet transform. To make the image fusion results of multi-spectral and high spatial resolution panchromatic image maintain the quality of the spectrum with high spatial resolution, this paper presents a fusion algorithm based on the second generation Curvelet transform remote sensing images. And the algorithm has been improved. The main research and work of this paper include:This paper presents a fusion algorithm Curvelet transform technique based on remote sensing images. The algorithm fusion rules take weighting coefficient in the low-frequency part and the larger absolute value method in the high-frequency part. The algorithm has greater resolution and contained more spectral information than traditional remote sensing image fusion algorithm. However, the algorithm is only suitable for the source image with high brightness and contrast and richer high-frequency part.This paper presents the algorithm combining the HIS and the Curvelet algorithm. This algorithm improved the Curvelet transform remote sensing image fusion algorithm based on local energy maximum. The average gradient of the fusion image is12.8401while the average gradient of the fusion image is12.9675after improved. The information entropy of the fusion image is6.6783while the information entropy of the fusion image is6.9338after improved. It improved the fused image clarity while the spectral content is richer.Then this paper improves the fusion rules of the Curvelet transform algorithm. The algorithm fusion rules take maximum value selection method in the low-frequency part and the regional feature comparison method in the high-frequency part. Experimental results show that the improved algorithm fusion effect the best. The average gradient of the fusion image reaches14.8153. And the information entropy of the fusion image reaches7.1350. The improved algorithm fusion has greatest resolution and contained the most spectral information compared with the Curvelet transform remote sensing image fusion algorithm based on the largest local energy and the Curvelet-HIS transform remote sensing image fusion.