A SAR image target feature extraction and recognition method based on kfda and svm

A technology of target features and recognition methods, applied in character and pattern recognition, instruments, computer components, etc., can solve the problems that the optimal classification surface is not the global optimum, cannot extract nonlinear features of images, and does not have the ability to classify , to achieve the effect of overcoming orientation sensitivity, good generalization, and low feature dimension

CN103824093BInactive Publication Date: 2017-10-13BEIHANG UNIV
4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Publication Date
2017-10-13
Estimated Expiration
Not applicable · inactive patent

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The present invention provides a SAR image target feature extraction and recognition method based on KFDA (Kernel Fisher Discriminant Analysis) and SVM (Support Vector Machine), comprising the following steps: performing training target samples of known categories and testing target samples of unknown categories Amplitude data normalization processing; using the KFDA criterion to perform feature extraction on the normalized known categories of training target samples and unknown categories of test target samples; The SVM classifier is trained to generate the optimal classification surface; finally, the features of the unknown category of test target samples extracted by the KFDA criterion are identified through the optimal classification surface; the present invention reduces the requirements for the preprocessing process and overcomes the The azimuth sensitivity of SAR images compresses the dimensionality of sample features and obtains a high target recognition rate, which has good generalization.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention belongs to the field of SAR image processing and pattern recognition, and relates to a SAR image target feature extraction and recognition method based on KFDA (Kernel Fisher Discriminant Analysis) and SVM (Support Vector Machine). Background technique

[0002] Synthetic Aperture Radar (SAR) is an active sensor using microwave perception. It can conduct all-weather and all-weather reconnaissance on targets or areas of interest. Penetration ability of ground objects. The so-called radar target recognition is based on the detection and positioning of the target by radar, and according to the radar echo signal of the target and the environment, the target features are extracted to realize the determination of the attribute, category or model of the target. With the continuous maturity of SAR imaging technology, target recognition based on SAR images has become more and more important.

[0003] In the target recognition process based on SAR...

Examples

Embodiment Construction

[0050] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0051] like figure 1 Shown, the specific implementation steps of the SAR image target feature extraction and recognition method based on KFDA and SVM of the present invention are as follows:

[0052] Step (1), normalize the amplitude data of the training target samples of the known category and the test target samples of the unknown category, and the normalization formula is:

[0053]

[0054] Among them, x is a vector representation of any known category of training target samples or unknown category of test target samples (that is, the image matrix is ​​arranged into a vector form by column), x Normalized It is a vector representation after normalization of the magnitude data of the corresponding training target samples of the known category or the testing target samples of the unknown category.

[0055] Step (2), using the KFDA criter...