Research On Color Remote Sensing Image Segmentation Method Based On Spatial Location Information

With the development of image segmentation technology and remote sensing technology,relevant researchers have invested a lot of energy and designed many image segmentation algorithms.However,because remote sensing images contain complex features and various features are interlaced with each other,a certain type of image The segmentation algorithm is difficult to apply to many different scenarios.For the segmentation of color remote sensing images,researchers have designed a variety of image segmentation algorithms based on the Iterative Conditional Mode(ICM)algorithm.However,there is still room for improvement in the quality of the initial segmentation results obtained by these algorithms,and the existing algorithms lack the Effective use of spatial location information in images.Based on the above two deficiencies,this paper proposes a color remote sensing image segmentation algorithm based on spatial location information.The research work of this article mainly includes the following aspects:Since the initial segmentation result of ICM algorithm still has room for improvement,this paper uses KFCM algorithm to replace the existing algorithm to obtain the initial segmentation result.The KFCM algorithm is an improvement of the FCM algorithm.It uses the distance in the feature space after Gaussian convolution instead of Euclidean distance to obtain higher segmentation accuracy.Before using the KFCM algorithm to process the image,the image is first bilaterally filtered.The use of bilateral filtering can not only reduce the interference of noise and smooth the image,but also effectively preserve the boundary information.Then input the initial segmentation result obtained by the above method into the ICM model and iteratively obtain the intermediate result of image segmentation.Aiming at the problem that the spatial location information in the image cannot be effectively used,this paper proposes a color remote sensing image segmentation algorithm based on the spatial location information.Since all types of features in the image are generally connected regions,a neighborhood is constructed for each pixel,and the segmentation result of the central pixel is corrected by using the spatial position information of the pixel in the neighborhood,and then the final image segmentation result is obtained.In the Google Earth part of the color remote sensing image data set of the United States and Japan,the algorithm proposed in this paper is used for comparison experiments with existing algorithms.The experimental results show that the quality of the initial labeling results obtained by the KFCM algorithm is higher;and the segmentation effect of the ICM algorithm based on spatial position information proposed in this paper is good,and the segmentation accuracy is above 85%,which is certain compared with the existing algorithms.To improve the segmentation accuracy of color remote sensing images,it is effective to use KFCM to initialize and introduce spatial position information for correction.