Study Of Target Fusion Identification Algorithm Based On Electronic And Optical Reconnaissance

Space-based fusion system has been an essential means of getting the power of controlling special information. Spaceborne electronic reconnaissance and aviation optical reconnaissance are its two important and widely-used instruments. Both have better complementation on location and identification precision, etc.The combination of them can ameliorate the capability of reconnaissance satellite, improve the ability of controlling information world widely. This thesis develops the theory and technology research of single target identification, identity hierarchy transition, and target association and fusion identification in the identification fusion employing spaceborne electronic and optical reconnaissance circumfusing the need of national defence.(l)The uncertain and its reason of battlefield environment are analyzed. Then Fuzzy Set theory is introduced to the single target identification algorithm. The result of emulation shows that fuzzy matching is prior to the template matching in the performance of identification, can efficiently improve the recognition rate and the ability of fault tolerance in strong noise environment. (2)According to demands to time and zone performance of configuration files between warship and its radar attachment, a new method of identity hierarchy transition induced by data mining was proposed. By using of algorithm for mining the fuzzy association rules, hidden configuration rules are extracted from huge amount of history identification results in database, radar-operator experience, military information, etc, which are better coincides with real condition. This algorithm is very meaningful to build a generic transition method of identity hierarchy. (3)The location and identification precision difference of between electronic and optical reconnaissance is analyzed firstly. Then a distance function to depict comparability between spatial position feature data and attribute evidences is constructed and the algorithm based on this function is proposed, in which the multi-target association problem is represented as a sort of constraint combinational sequencing optimization problem and the simulated annealing method is applied to it successfully. Numerical examples demonstrate that the effectiveness of the modified method is satisfactory. In the second part of chapter 3, evidential combination methods based on Dempster rule are analyzed firstly. (4)An evidential combination method based on imperfection of discrimination space and unreliability of evidence is built, and a target fusion identification algorithm is proposed by employing information of electronic and optical reconnaissance. Simulation results show that this method possesses obvious advantage in the case of target identification nicety and steadiness.Our identification fusion algorithm also has been used in the outsource data simulatedfusion system. It could fuse the information from sources of space born electronic and optical reconnaissance and identify.