Neural Network Based Research On Target Image Recognition

With the persistent development of Digit Image Processing and Artificial Intelligence, Pattern Recognition has made great progress. Acting as the stage of new technique, the modem war has more severe requirement to recognize aircraft targets. Thus, it is rather import to launch the research of air target recognition.This dissertation emphasize on the research of air target recognition, which can be divided into two parts: preprocessing and post classification and recognition of the image. This paper is structured as follows:Firstly, the source image is being preprocessed, including filtering, edge detection, iteration thresholding segmentation and binarization. In this thesis, Robert, Sobel, Prewitt, Gauss-Laplacian and Canny are used to detect edges. Then the image is iteration thresholding segmented and binaried.Secondly, moment feature of the image is extracted. In this paper, Hu moment feature is extracted, normalized and modified.Thirdly, moment feature, which is the input of artificial neural network and of Support Vector Machine (SVM) respectively, is used to recognize the category of the aircraft. In this paper, improved back-propagation network and Kohonen neural network, MLFNN network group and SVM classifier are proposed.Lastly, based on the methods mentioned, software system using VC Language is designed to test the theories. Results indicate that this system can recognize aircrafts and , the classifiers have high classification rate.