Information extracting method for remote sensing image continuous target field based on cluster analysis

Abstract

The invention relates to an information extracting method for a remote sensing image continuous target field based on cluster analysis, and belongs to the technical field of information extraction. The method comprises the following steps: carrying out feature analysis on original image data to find a maximum reference value or a minimum reference value of area character data of the target field to be extracted; carrying out threshold segmentation on the image data according to the maximum reference value or the minimum reference value; carrying out cluster analysis on the image data subjected to the threshold segmentation; rejecting a false target field from a target continuous field obtained by the cluster analysis; and carrying out mask processing on the image subjected to the false target field rejection and an original image to obtain a result image subjected to information extraction. With the technical scheme, the continuous field target can be identified and extracted more automatically; by using the cluster mode, on the basis of a target object, the image features endowed by the continuous field targets in different images are fully considered and the target which is really needed to be identified can be identified and extracted, therefore, compared with the prior art, the information extracting method has higher accuracy.

Description

technical field [0001] The invention belongs to the technical field of information extraction, and in particular relates to an information extraction method for continuous target fields of remote sensing images based on cluster analysis, so as to obtain information such as boundary, range, and distribution of feature fields. Background technique [0002] The common information extraction method of remote sensing images is classification, which can be divided into: supervised classification and unsupervised classification according to whether there is a training area; and common classifiers are based on maximum likelihood method, minimum distance method, vector machine classification, decision tree classification , object-oriented classification, and classification tools for neural networks. However, on the one hand, these classification schemes are difficult to achieve fully automatic information target extraction due to the large amount of manual participation; Because of ...

AI Technical Summary

Problems solved by technology

However, on the one hand, these classification schemes are difficult to achieve fully automatic information target extraction due to the large amount of manual participation; Because of the uniqueness of characteristic fields such as water areas on visible light images and oil spill areas on radar images), it is difficult to obtain satisfactory accuracy when identifying these continuous target fields

Images