The Research And Application Of The Feature Fusion And Extraction Algorithm In Multi-resource Images

With a wide range of theoretical and practical application of the significance, multi-source image fusion and extraction technology is widely applied in a variety of fields such as military surveillance, computer vision, medical diagnosis and remote sensing. The paper includes two parts, which is the fusion and extraction of regional features and linear features in multi-source images.The connected region is composed of adjacent pixels in the space by region-extraction based on some consistency in the criteria. And region-extraction can extract image features based on region, which is very effective for the target description; while straight lines can describe a number of man-made object, such as building, the airport and roads and so on, and the line is a very important primitive in scene matching, stereo matching and sequence analysis. Therefore, the paper puts forward some analysts approach from the regional features and linear features extraction. For the extraction of region issues, at first, the paper analyses the differences, and advantages and disadvantages of the region segmentation and extraction problem based on Euclidean distance and Mahalanobis distance, and then propose a regional extraction algorithm combined of fuzzy C-means clustering (FCM); As for the extraction of linear features problem, the paper puts forward a linear fusion and extraction algorithm based on the phase group algorithm, which is applied to road extraction problem of remote sensing images.All the algorithms paper presents are applied to multi-source images data. Experimental results show the effectiveness of the algorithms.