Research On Matching Algorithm Of Drogue Infrared Images

Air refueling technology plays an increasingly important role in modern warfare and has becomea national military air multiplier. When air refueling is occurred at night or in harsh weatherconditions, the collected visible imaging of visual navigation system is fuzzy and dim, and unable toachieve the docking of the refueling drogue and probe. But the image formed by the reflected infraredcan make the invisible objects visible. The thesis researches on the matching of drogue infrared imageand achieves the air docking functions at night.Firstly, the thesis describes the mechanism of infrared image, analyzes the composition ofpassive and active infrared imaging system and their difference, and analyzes the characteristics ofinfrared images in multiple scenes. According to the characteristics of infrared image, thepre-processing of infrared images are studied which can improve the contrast of the image andeliminate background noise.Secondly, the matching algorithms based on gray of infrared image are studied. Some similaritymeasures are compared and analyzed. A compound fast matching algorithm based on the idea ofmulti-resolution is proposed to achieve the coarse matching by lowest resolution image and finematching by highest resolution image. The results show that the algorithm has a higher matchingspeed and matching accuracy.Then, the thesis studies the feature-based infrared image matching algorithm. Firstly, severalfeature-extracting algorithms are described in detail, A combined Harris and NCC algorithm is used tomatch feature points, experimental results show that the matching algorithm has a high error rate.Then the feature-extracting feature-describing and feature-matching of SIFT algorithm are studied，then a HDRSIFT algorithm is proposed, which improves SIFT algorithm from two aspects：reducingthe feature points and dimension. In the experimental stage, a variety of experiments analysiscertificates that, compared to SIFT algorithm, HDRSIFT algorithm has a better performance in termsof the matching efficiency and accuracy.Finally, the thesis constructs a drogue infrared image acquisition system, a double droguedetecting process based on gray matching and feature matching is developed and designed, Theresults of this study show that the algorithm studied by the thesis can detect infrared images drogueeffectively，and has a better detection stability and time efficiency.