The Recognition Of Study Passivity Target

The dissertation studied the method for extracting and recognizing of under-water target character based on the existing methods. The author of this dissertation introduced Information Fusion (IF) Technology to the paper. At last, I conclude a way to increase the recognition ratio of the under-water target.The method of this dissertation is to distinguish the property of the under-water target by eradiate noise. Eradiate noise recognition obtains feature information of movement under-water target, which is produced by passivity sonar, and then decides its category by referring to a priori knowledge. The dissertation tries to combine signal processing and information fusion with the application background of under-water recognition so as to improve the reliability and accuracy of eradiate noise recognition system. And based on this, the author designed a software system to make the process of recognition visual. In the course of programming, the author made use of the technique of blending C++ Builder and Matlab to improve code efficiency.According to the principle of mode recognition, this paper can be mainly divided into these parts: Feature extraction, classification and recognition, experimental study.The process of feature extraction is to transform the eradiate noise signal to different feature space and extract the feature vectors that reflect the category of the input sample. The extracted features are the input modes to the classifier. Through the inspecting, the author selected three methods of eradiate noise's feature extraction. The methods include the pedigree, E-Filter, fractal. In those methods, the Pedigree is based on frequency field, the E-Filter is based on the energy field and the fractal is based on time field. Above all, the system fuse theeradiate noise's different sides' feature and offer strong elements for the next process: classifying.It is very important to study the classifier in this dissertation. It is the key whether IF can be made use of in the field of eradiate noise recognition efficiently. The definition of classifier is that the classifier can decide the category of current input sample according to its pattern. The classifier that this dissertation introduced is the MLP classifier design under the condition of ANN. It can improve the system's performance if use three MLP fusion classifiers to make decision fusion.At last, it is the part of dealing with the data is gained in experimentation. We have made the experimentations in Songhua Lake. In the paper, I analyze and deal with the primal data of eradiate noise to get the ratio of recognition and make a conclusion that the feature extraction arithmetic and classifier arithmetic are efficient.