Research And Implementation Of Remote Sensing Image Scaling Technology Based On Content Perception

With the progress of aerospace technology,how to achieve the remote sensing image in the sensitive area without distortion of the effective scaling has become a very important research hot spot.Airports and bridges play an important role in transportation and have been the important goal of national defense and military strikes.It is important to observe the details of these two objects in the remote sensing image from the global point of view.It is an important basis for military strikes and research.In this paper,it proposes a method to extract the sensitive features of objects and combine them with the general characteristics of the objects to identify sensitive objects in multi-spectral images.And a remote sensing image scaling method based on a triangular mesh and fisheye transformation is proposed to achieve the optimal scaling of the sensitive region.Multispectral images have many classification features because of their different acquisition methods.The extraction of these features can effectively identify different object regions in the image.In this paper,a sensitivity-based sensitive region discrimination method is proposed to extract the texture region,the color moment feature and the area feature of the extracted sensitive object potential region,and then extract the object region sensitivity feature in the context of the original image,and finally,through the AdaBoost SVM classifier based on the classification of the potential areas to determine the specific location of sensitive areas,and identified in the source image.This method is applied to multi-spectral image airport and bridge-sensitive target recognition.The experimental results show that the average recognition accuracy of the method is higher than the airport identification method which based on the airport runway straight line characteristics and the bridge identification method which based on the spectral differences between the bridge and water.The average accuracy of the airport identification is 92%.The average accuracy of the bridge identification is 90%.At present,there are very few researches on remote sensing image scaling technology.In this paper,an image scaling method combined with a triangular mesh and a fisheye transformation is proposed.For the sensitive regions in the identified multi-spectral images,the triangular meshes are divided into regions,and the zoom based on triangular mesh is implemented in the regions and hold the contents of the sensitive part of the area.The part of the whole image except the sensitive area is compressed based on the fisheye transformation method associated with the change in the sensitive area scaling process.The experimental results show that the distortion most occurs in the part of the insensitive region,and the distortion inside the sensitive area is very small,and the zoom effect is more obvious than the existing improved line crop scaling method,Triangular Mesh and Fisheye Transform methods.