Method and system for estimating number of classes of ground features in remote-sensing image

Abstract

The invention provides a method and system for estimating the number of classes of ground features in a remote-sensing image, and the method comprises the steps: segmenting a to-be-segmented remote-sensing image through employing a fuzzy clustering method, and obtaining an optimal fuzzy membership degree and an optimal fuzzy clustering center; calculating the initial information entropy of the classes of ground features and the upper limit of information entropy: carrying out the splitting operation if the initial information entropy is greater than the upper limit of information entropy, or else, calculating the Euclidean distance between the optimal fuzzy clustering centers; determining that two classes of ground features are similar if the initial information entropy is less than a given threshold value and carrying out the merging operation, or else, obtaining an estimation result of the number of classes of ground features and a final segmentation result. The method can combine with a conventional image segmentation method, and effectively estimates the number of classes of ground features in the remote-sensing image. Based on the information entropy, the method measures the characteristics of information quantity in the classes of the ground features, defines a splitting condition, and describes the difference among different classes of ground features through employing the Euclidean distance. The method also defines a merging condition, can split a region, which cannot be split effectively in a clustering algorithm, in the splitting operation, and improves the segmentation precision.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a method and system for estimating the category number of ground objects in remote sensing images. Background technique [0002] Determining the number of categories in an image is an important task in the process of image segmentation. The current category estimation methods are mainly divided into two categories. One is to manually determine the number of categories, but this method is not suitable for high-resolution large-scale images with obvious details; The other is to automatically determine the number of categories through the design method. At present, these methods are designed with the goal of automatically determining the number of image categories and region segmentation at the same time. This design method is relatively simple in principle, but it is highly targeted and not universal. In addition, there are many thresholds to control the change of the numb...

AI Technical Summary

Problems solved by technology

[0002] Determining the number of categories in an image is an important task in the process of image segmentation. The current category estimation methods are mainly divided into two categories. One is to manually determine the number of categories, but this method is not suitable for high-resolution large-scale images with obvious details; The other is to automatically determine the number of categories through the design method. At present, these methods are designed with the goal of automatically determining the number of image categories and region segmentation at the same time. The principle of this design method is relatively simple, but it is highly targeted and not universal. In addition, there are many thresholds for controlling the change of the number of classes, and the interaction between multiple thresholds makes it difficult to obtain ideal estimation results, and requires a large time cost, it is difficult to process large data, and its practicability is not strong

Images