Infrared And Visible Images Targeted Research And Realization Of Multi - Mode Tracking Technology

The design and implementation of multimodel (optical and infrared) image tracking system are discussed in this paper. Based on the domestic and abroad advanced technique, this paper introduces the theories on correlation/contrast tracking, and reviews the related software/hardware implementation techniques from the engineering perspective. We also present the implementation and the performance of multimodel image tracking systems. Based on an in-depth discussion on hybrid techniques for multimodel image tracking, we propose new methods with relatively highly practical value, which have been used successfully in practical engineering project. In this paper, considering of the differences of objects and background in the optical and infrared image, the correlation and contrast tracking techniques will be used in optical or infrared image tracking, and then based on the confidence figure of the two separate tracking system, the output of multimodel tracking system is decided. We discuss each component according to the system structure, and introduce software/ hardware of the multiple DSPs. This mainly involves two aspects: image segmentation and image matching. The former introduces a fast technique for extracting shape centroid and determining segmentation thresholds on the specified cases. The latter focuses on the techniques for model extraction, model matching and related criterions. The implementation and performance of this proposed multimodel image tracking system are demonstrated with four chapters. The results presented show our methods improved accuracy and efficiency. Finally, we discuss the limitations of the proposed system and the future work.