| Infra-red search and tracking (IRST) system is a kind of passive infra-red detectionsystems, and is one of the important units in defense system for modern war, which hasadvantages of good concealing, high angular resolution and high anti-EM interferencecapability. With the development and application of optoelectronic countermeasurestechnology, the flight velocity of fighters and missiles is significantly increased,operational modes for missiles show polymorphism(for example, one missile with multiwarhead, true or false missiles firing simultaneously and multi-azimuth group hair),and the operational war environment becomes more complicated. To meet these factorsrequirements, the IRST system must be improved in response speed,field of view andspatial resolution,etal. In this thesis, omni-directional IRST system based on spatialmultiplexing IRFPA staring imaging technology is put forward, which can meet aboverequirements. The proposed information processing algorithm and hardware realizationis researched for the omni-directional IRST system in the thesis.System structure and working principle for omni-directional IRST system basedIRFPA staring imaging technology are analyzed, then key technologies in informationprocessing algorithms, such as, infrared image preprocessing, weak target detection andtracking and information fusion for multi sensors, are further studied. Informationprocessing algorithms for the omni-directional IRST system are proposed. Thecorrectness and efficiency for above algorithms are verified with theory analysis andsimulation verification. Finally, hardware for realizing these algorithms is also provided.Main works in this thesis are list as follow:1. Based on analysis on the generation mechanism of non-uniformity in IRFPAstaring imaging algorithm and shortage for common non-uniformity correctionalgorithms, non-uniformity correction algorithm with multi-resolution wavelettransforms is researched. The gain and bias parameters of non-uniformity arecorrected by using this algorithm. More important, its efficiency is verified withexperimental results.2. According to complex background and interference of man-made targets inworking environment for IRST system, background suppression algorithm based on allphase non-subsampled contourlet transform is put forward. Combined phasenon-subsampled Contourlet transform with fuzzy theory and total variation model isrealized to improve the adaptability of the algorithm. Simulation results show that ahigh signal-to-noise ratio is achieved. This algorithm has advantages in simple structure, strong robustness and good real-time performance.3. By using auto-thresholding segmentation algorithm and inter-frame relatedsequence image detecting algorithm, weak target accurate detection is realized,and theperformance of target detection algorithms are evaluated by using the detectionprobability, false alarm probability and ROC curve. In order to realize In order to realizetarget tracking with many kinds of moving state, the maneuvering target tracking algorithmbased on the interacting multiple model Kalman-Gauss partice filter is proposed. Thisalgorithm has the characteristics of high precision,strong adaptability and goodreal-time, and has a good engineering application prospect.4. Information fusion technologies in omni directional IRST system areinvestigated with target tracking fusion and image mosaic algorithm. By adoptingdistributed tracking fusion structure, tracking association and fusion algorithm isproposed. To realize the combination small view field images originated from IRFPAsensors into one omni directional image, fast image mosaic algorithm based onimproved SURF feature detection, RANSAC feature points matching and weightedaverage pixel fusion is realized.5. To realize the proposed information processing algorithm in omni directionalIRST system, high speed parallel image processing hardware platform based on FPGAand multi DSP is obtained. RF5software framework for DSP/BIOS real-time operatingsystem is also achieved in thesis. The hardware platform has advantages, not only inhigh precision and real-time, but also in well stability, easy extension and co nvenientmaintenance.In a word, image information processing algorithm of omni-directional IRSTsystem and the software&hardware platform for real-time realized proposed algorithmin the thesis can satisfy the performance requirements of the research project. Theseresearch results explore an effective way to research high performance omni-directionalIRST system, and provide a certain foundation for the project implementation of theomni-directional IRST system. |
PDF Full Text Request