Research On Multi-temporal High Resolution Remote Sensing Image Change Detection Algorithm

With the rapid development of remote sensing technology,remote sensing acquisition methods are increasingly diversified,providing important data support for the practice and application of remote sensing theory.Remote sensing image change detection extracts the change information of the target area by comparing and analyzing the remote sensing image features of the same area at different times.It is an important direction of remote sensing applications and has been applied in various fields.Change detection plays an important role in urban planning,environmental monitoring,and disaster prevention.It is widely used in civil and military fields.By monitoring the characteristics of the Earth's surface,important information that is valuable to human beings can be obtained.Image change detection technology has gradually become a research hotspots in the field of remote sensing.In recent years,a variety of remote sensing image change detection methods were proposed.According to whether a priori knowledge is needed,it is divided into two types: supervised change detection and unsupervised change detection.It is necessary to select an appropriate method according to the application requirements to improve the performance of the change detection.In this paper,the SAR remote sensing image and the high-resolution reef image change detection algorithm are studied.The corresponding solutions are proposed according to their respective characteristics,and the effectiveness of the algorithm is verified by experiments.The main innovations are as follows:Aiming at the problem that the synthetic aperture radar remote sensing image is seriously interfered by the speckle noise and the detail information of the changing region is preserved,an image change detection algorithm based on the saliency-guided sparse automatic encoder is proposed.The change region in the initial difference image has local consistency and global prominence.By detecting the similar change region by significant region detection,the speckle noise outside the region can be eliminated.In order to improve the reliability of training samples,hierarchical FCM clustering is used to automatically generate training samples.In addition,in order to increase the discrimination between the changed region and the unchanged region,a sparse automatic encoder is used to extract the variation feature,and a change detection map is generated.Experiments were performed using 4 sets of SAR images,and the results show that the proposed algorithm can effectively reduce the influence of speckle noise on detection accuracy.The extraction of the change region is more complete and detailed,and the error detection rate is greatly reduced,which is more suitable for SAR remote sensing image change detection.Aiming at the problem that the remote sensing image of high resolution reef is greatly influenced by the external environment and the change region is difficult to extract,a change detection algorithm based on target segmentation and PCA-Kmeans is proposed.Since different phase images may be disturbed by weather,clouds,seawater,etc.,the target segmentation algorithm may be used to extract the target of interest and highlight the changed region.Principal component analysis and kmeans clustering were used to reduce the influence of isolated pixels,and the change information was extracted to obtain the difference image.The experiment uses four sets of image data of the reef area.The experiment proves that the algorithm can eliminate the external interference well,improve the detection accuracy of the change,and have a good detection effect on the area of the island reef.In summary,the two high-resolution remote sensing image change detection algorithms proposed in this paper play a very good role in the detection of change regions,and the visual effects are better,it can improve the accuracy of the test results.