Research On Extended Target Stable Tracking Technology In Sky Background

Target tracking is a key technology of the optoelectronic measurement system.The extended target stability tracking technology in the sky background is one of the core issues in target tracking,which is widely used in practical scenarios.Its research has important implications for the technology development in the target tracking field and also has important practical engineering value.Although some classical algorithms have been proposed at home and abroad,there are still many problems in practical application.To improve tracking stability of the extended target,an algorithm that is universally adaptable and robust to these cases needs to be researched.In this paper,the principle and the defects of the existing algorithms are analyzed firstly.Then,the problems that need to be solved in the actual project of the extended targets stable tracking in the sky background are studied in depth.The new and high adaptability algorithms that can achieve stable tracking of the extended targets are proposed,the main research contents are:A detailed study on the characteristics of the extended target image under the sky background is given,and the problems to be solved and the main difficulties existing in the extended target tracking under the sky background are analyzed and summarized.The common image features of the extended target are studied,and the existing problems of the extended target tracking regular method based on the local features are analyzed.The principle,advantages and disadvantages of generalized Hough transform are studied in depth,and the extended target tracking method based on generalized Hough transform is discussed deeply.Experiments show that this algorithm is of certain value to the stability tracking of extended targets.To solve the drift problem caused by the rotation and scale changes and the target region beyond the field of view in the extended target tracking process,the advantages and disadvantages of the extended target tracking method based on the generalized Hough transform are analyzed.An extended target tracking point extraction method based on ORB feature point matching and generalized Hough transform is proposed.Experimental results show that the proposed algorithm can reduce the drift to a smaller extent for a smaller degree of rotation and the variation of the scale and the target partial area exceeds the field of view.On this basis,the principle of GMS algorithm is deeply studied,and an extended target tracking point extraction method based on GMS and generalized Hough transform is proposed.Experiments show that the algorithm is more adaptable to a larger degree of rotation and scale changes.Aiming at the drift and jitter of tracking point caused by illumination changes,cloud interference and self-occlusion caused by target attitude changes in the extended target tracking process,an extended target stable tracking point extraction method based on local and global coupling is proposed.Experiments show that the proposed algorithm has high robustness to changes of target rotation,scale,illumination,self-occlusion,cloud interference and target field beyond the field of view,and the extracted tracking points have sub-pixel accuracy and greatly improve the stability of the extended target tracking.Aiming at the fact that the extended target flight sequence in the sky background in the actual project becomes blur from the clear or becomes clear from the blur,a self-adaptive extended target stable tracking point extraction method is designed.The degree of ambiguity of the target is measured by the evaluation of the No-Reference Structural Sharpness.Then selecting the corresponding algorithm to extract the tracking point tracking target.Experimental results show that this method can meet the actual needs better.