Research On Infrared Image Detection Technology Of Unmanned Aerial Vehicle In Low Altitude Airspace

The fast and accurate detection of infrared dim and small targets is the premise of following tracking and analysis of the target by infrared monitoring and warning system.At the same time,the realization of the detection of unmanned aerial vehicles with low space is a necessary measure to the current unmanned aerial vehicle supervision and airspace monitoring,The all-weather and all day characteristics of infrared image sensors are better than visible light through the fog,snow and atmosphere,which makes the infrared sensor a good choice.This paper mainly studies the detection of low altitude and long distance small targets in the field of vision based on infrared detection technology under complex scenes.Through the improvement of soft morphological filtering and the separation of the background,the image is enhanced with a fractional differential operator and a significant detection method,and a complete dim target is extracted to achieve the purpose of effective detection.The main contents of this paper are as follows: This paper briefly introduces the application background and research significance of infrared small target detection technology,and analyzes the advantages and disadvantages of some common algorithms based on the thinking of tracking before detection in recent years,combined with the current research status at home and abroad as well.1.The soft morphological operators are improved from two aspects.Compared with the traditional morphological filtering,the improved algorithm has the ability to suppress the background.Compared with the soft morphological filtering,the improved top-hat operation can reduce the running time while it achieves the same background inhibition effect.2.In the stage of target extraction,a significant detection method is introduced with fractional differential.Firstly,the fractional differential is used to enhance the target,the target can be highlighted and the noise can be suppressed to a certain extent,and the significance of the target is increased.Then a significant detection algorithm can be used to better extract the target.3.In the threshold selection stage,two improvements are made to the commonly used threshold selection method in the threshold segmentation stage,which makes theuse of improved threshold segmentation method to extract two targets more accurately when there are two targets with different significant situations in the image.4.After extracting the target in the single frame target processing stage,in order to further reduce the false alarm rate,the pipe diameter selection rule is improved on the basis of the original pipeline filtering algorithm,and the area information of the target is further utilized.Based on the spatiotemporal continuity of the moving target(the continuity of the target position and the inter frame correlation),the multi frame correlation is taken advantage of.The method can eliminate the isolated noise point to a certain extent.The experiment shows that the algorithm used in this paper has high detection rate and low false alarm rate in the three scenarios.In general,the detection effect is better than the other four groups of contrast algorithms.