A Study On Track Correlation Method Based On Time Sequence SVM Information Fusion For Target

Based on many related subjects and theory, information fusion is a data processing course that automatically analyzes and synthesizes the data gathered from various sources under time rules. If it was applied to the radar target recognition, there will be far-reaching research value to the war in the future.This information fusion discussed in this paper aims at the problem of target track correlation of radar. The target track correlation of radar is a multi-classification problem, and it is difficult to obtain prior knowledge and sample data in advance. In term of observing time sequence, this problem can be divided into several dyadic problems. The reasoning method based support vector machine can solve dyadic classified problem very well.In the basis of the analysis of information fusion process and the methods used for radar target recognition, the paper will explain the reasoning method of support vector machine and introduce it into the problem of the radar target track correlation, and improve the reasoning method based support vector machine (SVM) during combining it with the time sequence. The paper also expanded sample data set gradually according to the observing time sequence, and made the selection and classification of the sample data at the same time. Time sequence support vector machine (SVMt) was brought out which presents a method of data classification to solve the problem of track correlation. In this paper, the theory of SVMt as well as the criterions and algorithms of data classification is also described.Finally, the paper proved the validity and adaptability of SVMt through simulation experiments.