Research On Image Fusion Algorithm Based On Feature Difference Segmentation

In the field of image processing,image fusion occupies an important position and is applied by various technology industries.Image fusion technology uses a specific algorithm to comb the multi-source image information,and fuses the complementary information to generate a more comprehensive and reliable image.In this paper,the fusion algorithm of multi-spectral image and panchromatic image is deeply studied.Aiming at solving the problems existing in multi-resolution analysis-based fusion method,an image fusion algorithm based on feature segmentation is proposed in this article.The specific work is as follows:First,research the background and significance of image fusion technology,and understands the current status and development trend of current image fusion technology.Then the existing remote sensing image fusion methods and fusion evaluation criteria are summarized,and the principle and process of multi-resolution analysis-based fusion method are deeply studied.Analyze the advantages and disadvantages of these methods via simulation experiment.In order to make up for the shortcomings of the above methods,the idea of image segmentation is introduced into the fusion.So this paper do a deeply research on image segmentation theory and some common methods.Aiming at solving the shortcomings of traditional fuzzy C-means Clustering(FCM)algorithm,a new FCM algorithm based on neighborhood constraint is proposed.The algorithm fully considers the neighborhood information of the pixel points,and effectively solves the problem of the selection of the initial clustering center and the problem of being sensltive to noise and isolated points.So the final segmentation effect is more suitable for the next image fusion.Based on the research of multi-resolution analysis and image segmentation,an image fusion algorithm based on feature segmentation is proposed.The algorithm performs Dual-Tree Complex Wavelet Transform(DT-CWT)on the source image to separate the spectral information and spatial detail information primarily.Then the new FCM algorithm based on neighborhood constraint and the edge detection algorithm are used to segment the low frequency and high frequency images respectively.After segmentation,the low-frequency region is fused by the rule based on the regional significance metric,and the high-frequency region is fused by the rule based on the edge feature.Finally,the DT-CWT inverse transform is performed on the low frequency and high frequency to reconstruct the fused image.The simulation comparison experiments show that the proposed method effectively improves the spectral information and spatial resolution of the fused image,and its subjective and objective evaluations have a good performance.