抚仙湖太阳观测和研究基地知识库

This paper first studies motion-related data driven tracking methods and analyzes their insufficiency and applications, then introduces the idea of Mean Shift and constructs kernel probability density based target model in accordance with statistical characteristics of infrared dim and small target in terms of statistical characteristics based difference between the infrared dim and small target and noise, and finally explores key problems such as target template extracting from tracking infrared dim and small target, tracking location determination and target model updating. The experiment results have shown that this motion-related based tracking method incorporating target gray scale statistical characteristics achieves effective combination of two tracking patterns by integrating advantages of both and thus significantly improves accuracy of tracking infrared dim and small target.