Remote Sensing Image Segmentation Based On Saliency Map Extraction

With the rapid development of remote sensing technology,remote sensing images include more and more ground information,and the detection of objects of interest in remote sensing images has become one of the research hotspots in remote sensing,mapping,military and other applications at home and abroad.The detection of various types of targets in remote sensing images taken by satellites has become an important part of image analysis.As an important visual feature in image processing,the saliency of an image can show the degree of attention of human vision to different regions in a certain scene.With the rapid development of modern information technology and the needs of modern society and war,the image saliency detection is also appearing more and more in the field of image processing,such as image edge enhancement,image target partitioning,image compression,image feature extraction.etc.In this paper,the purpose of "segmentation of target of interest in remote sensing image" is that after consulting a large number of domestic and foreign literatures,it is found that the target detection technology in existing remote sensing images has poor information description ability,poor robustness,and no Technical problems such as uniform application of templates,poor real-time performance and long time-consuming.Based on the learning and research of the explicit graph extraction algorithm,based on the ITTI algorithm model,an improved graph extraction algorithm based on the ITTI algorithm is proposed to overcome most of the existing problems.By replacing the color image with the gray image,and then extracting the gray and the direction features by wavelet transform,combined with the idea of??GBVS algorithm model,the image nodes and connection weights are constructed based on the extracted image features.The feature map is then fused,and finally the saliency map is generated.Then,the image threshold segmentation technique and the image region labeling technique are used to detect the target in the remote sensing image.The experimental results show that the improved algorithm achieves excellent segmentation results,and has excellent segmentation results compared with the classical algorithm and the current popular target segmentation algorithm.