SAR image classification method based on spm and depth incremental svm

Abstract

The invention discloses a SAR image classification method based on SPM and depth incremental SVM. The steps are: (1) input SAR image; (2) extract dense SIFT feature of SAR image; (3) construct dictionary; (4) sparse coding; (5) spatial pyramid pooling; (6) calculate normalized feature; (7) Build an incremental training set; (8) Initialize the depth incremental support vector machine; (9) Calculate the initial classification accuracy of the test sample; (10) Update the depth incremental support vector machine; (11) Calculate the test sample classification accuracy. The invention can effectively extract the spatial information of the image, combines the advantages of deep learning and incremental learning, and has the advantages of improving the classification accuracy of SAR images and reducing training time.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a synthetic aperture of a spatial pyramid matching (Spatial Pyramid Matching, SPM) model and a depth incremental Support Vector Machine (Support Vector Machine, SVM) in the technical field of synthetic aperture radar image target classification Radar (Synthetic Aperture Radar, SAR) image classification method. The invention can be used for target classification and recognition of SAR images. Background technique [0002] Synthetic Aperture Radar (SAR) has the characteristics of all-weather, long-distance, extremely strong penetrating power and high resolution, and is widely used in the national economy and military fields. In the face of the ever-increasing SAR image data collection capabilities, how to accurately and quickly understand and identify these images has attracted more and more attention and attention. [0003] Xidian University disclosed a method bas...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that the feature extraction part is not targeted, so that the learned features cannot represent the original image well, resulting in a low recognition rate

However, the disadvantage of this method is that when the number of training samples changes, all training samples must be recombined, and the computational complexity increases significantly, resulting in long training time, low recognition efficiency, and poor real-time performance.

Images