Design And Implementation Of Target Recognition System Based On Cooperative Multi-sensor

The single sensor provides limited information, while multi-sensor cooperation could integrate each information. Therefore, the target recognition system based on multi-sensor cooperation is mainly researched in this thesis, which is significant in theory and practical in application to research on how to make use of the complement information of the sensors, and realize all day, all-round, accurate object recognition. To realize the target recognition system based on cooperative multi-sensor, we study the following contents:1) In this thesis, four kinds of characteristic are used in the target recognition system, which are Hu moment, affine invariant moment, wavelet moment and gray level co-occurrence matrix. Each kind of characteristic are analyzed in detail, and The calculation process of the feature is displayed. Finally, experiments have proved that these four characteristics can be divided between class and the similar classes are invariant.2) A feature fusion method based on PCA-KCCA is proposed on foundation of PCA algorithm. Use PCA to reduce the dimension of the data in the first, and then in a low dimensional space using CCA method for solving integration features. Through experiments has verified the feasibility of this method, and get a good recognition rate.3) The paper presents a fuzzy K neighbors-based fuzzy support vector machine method based on the existing studies. The method firstly calculates the mean of each sample and obtains the center of each class, and then calculates the distance between the samples and their center, reckons the initial membership of samples. The K neighbors of each sample were found out, according to which the fuzzy K membership of each sample was calculated, by mixing the initial membership and fuzzy K neighbors membership at certain proportion, the final membership was got, which is applied in fuzzy support vector machine. Then with the results of experiment, the performance of traditional SVM,Fuzzy SVM, Fuzzy SVM based on Fuzzy K neighbor nearest algorithm is compared.4) On the basis of the above algorithm, this paper presents a target recognition system based on multi sensor cooperation, which includes data acquisition module, a feature extraction module, multi-sensor cooperative module, database management module, recognition model training module, target recognition and target positioning module. C++ object-oriented language is used for each module of the system design and implementation, then all modules are integrated together, finally achieved the target recognition system based on multi sensor cooperation. The experiments proved that the recognition system has the good stability and the recognition speed and accuracy, so the system has very practical value.