Recognition method, based on deep belief network, of three-dimensional SAR images

Abstract

The invention provides a recognition method, based on deep belief network, of three-dimensional SAR images. The method comprises the following steps: firstly establishing a simulation sample bank of the three-dimensional SAR images, performing projection to different azimuthal angles and pitch angles through one or a small quantity of objective three-dimensional SAR images, so as to obtain a plurality of two-dimensional SAR images, ensuring that the small quantity of obtained three-dimensional SAR images are converted into two-dimensional images, and performing recognition through a two-dimensional image recognition method, and the method can greatly reduce the cost, and reduce the time for acquiring SAR imaging. According to the method, a splicing crossover verification method is proposed, and the deep belief network is improved, so that the deep belief network can automatically adjust parameters, self optimization of parameters is realized, the occurrence of over-fitting learning state and under-fitting learning state is effectively avoided, advance features of sample data can be accurately learnt, a better recognition result is obtained for the deep belief network, the complexity of manual setting of parameters is eliminated, and the recognition efficiency is improved.

Description

Technical field: [0001] The technical invention belongs to the technical field of radar, and in particular relates to the technical field of synthetic aperture radar (SAR) imaging. Background technique: [0002] Synthetic Aperture Radar (SAR) can detect and monitor the target area of ​​interest all-weather and all-weather without being restricted by natural conditions, and has been widely used in both civil and military fields. [0003] SAR image target recognition is the application of pattern recognition and artificial intelligence in SAR system, and its process can be divided into training sample stage and test sample stage. In the training phase, the SAR image of the training target is preprocessed first, including denoising, segmentation, contrast enhancement, etc., and then the stable and distinctive features of the target in the SAR image are extracted, and meaningful features are found from them. These features are used to design target recognition classifiers; in t...

AI Technical Summary

Problems solved by technology

And, on the recognition method, the present invention proposes the fusion cross-validation method to improve the deep confidence network so that it can automatically adjust the parameters and realize the self-optimization of the parameters. The advantages of this method are: 1) solve the problem of manual parameter setting The method is inefficient, and often fails to achieve the optimal parameter setting; 2) Using the cross-validation method can obtain a higher accuracy rate, and can effectively avoid the occurrence of over-fitting learning and under-fitting learning states; 3) The optimized parameters can accurately learn the advanced characteristics of the sample data, enabling the deep belief network to obtain better recognition results

Images