Research On Removal Of Thermal Blurring Around Infrared Cooperative Targets In Vision-based UAV Landing

Vision-based UAV autonomous landing technology requires co-operation from the ground targets with a specific shape to estimate the information of unmanned aerial vehicles, such as height, landing flight attitude information. The research group around to meet the requirements of under the UAV autonomous landing proposed a new method with ground-based infrared and airborne infrared vision system for UAV landing. However, the experiment found that, due to the infrared co-operative target at a high temperature state, it's easy to form blur around the co-operation target and this ambiguity could do a greater impact of the UAV position and attitude of the parameter estimation accuracy. To solve this problem, this paper started the research work from both the objectives of cooperation's pre-design and post-IR blurred image restoration areas.First of all, this paper analyzed the formation principle of thermal blurring in the angle of the infrared imaging mechanism. After a detailed analysis infrared thermal blurring from thermodynamic principle, the paper derived the main factors which are formation of infrared thermal blurring, and used Fluent simulation software to simulate the formation of thermal blurring. At the same time, the paper optimized the design of cooperation target to reduce thermal-blurring at the beginning. Experiments show that the optimum design of the cooperation objectives'thermal blurring has significantly improved compared with the old ones.Secondly, due to the theory of thermal blurring is different with the general principles of image blurring. In this paper, access to a large number of existing literatures chose the transparency-based deblur methods to apply to this context. For the original method can not be realized of automatically extracting a transparent region's problems, the paper used the method combining with simulation and experimental verification to set up the largest proportion of blur area, and also used improved Hengjing thinning algorithm to extract the blur area, and then the use of transparency-based on image restoration algorithm to restore images. Experiments show that the improved method has a good result.Then, through the study the law of actual image of the blur region with gray transformation curve and the temperature field in the simulation experiment to explore another way to remove thermal blur approach. Experiments show that this method can remove the thermal blur in a certain extent.In addition, since the background temperature is not high and the low rate of infrared emission and other reasons, the background in infrared images show a lower energy, while the thermal blur area in the image is not as apparent as cooperation targets. While the singular value decomposition (SVD) and image reconstruction with the extracted image of the role of different energy components. Therefore, we researched the removal of thermal blur method with SVD-based. The feature vectors that ranked by the size of characteristics of value with the SVD method to determine suitable number for reconstruction to deblur the image. Experiments show that the method has good treatment effect.Finally, through comparative experiments, combined with the background requirements that automatically extract the subject, the final choice is the improved method of transparency-based deblur algorithm as the subject of the final thermal blur recovery method.