Research on Recognition Technology of Low-Altitude Low- Speed and Small-Target Aircraft Based on Block Diagonal Feature

With the development of Low-altitude Low-Speed and Small-Target Aircraft (hereinafter referred to as LLS Aircraft), the number of dangerous accidents caused by them increases sharply. There is an urgent need to carry out research on the defense technology of LLS Aircraft. In the defense process of LLS Aircraft, the most important part is to recognize LLS Aircraft. Only on the basis of accurate recognition can we further achieve the accurate tracking, interception and disposal of different LLS Aircraft. Proposed in this paper, we study the LLS Aircraft recognition technology based on block-diagonal feature. We firstly extract the Histogram of Oriented Gradients (HOG) feature of LLS Aircraft images. Then the HOG feature is diagonalized by using low-rank recovery technology. The low-rank feature representation of LLS Aircraft is constructed by introducing a block-diagonal sparse regular term to increase the discrimination of LLS Aircraft feature. Based on the block-diagonal HOG feature, we use Support Vector Machine (SVM) to classify the LLS Aircraft, to enhance the accuracy of LLS Aircraft recognition.