Research On Infrared Small Target Detection Technology Under Complex Background

Infrared small target detection technology is one of the important research contents in the field of image processing,and it is widely used in the fields of infrared early warning and precision guidance.With the rapid development of UAV technology,infrared small target detection technology,which is the core of anti-UAV technology,needs to be improved quickly.Because of the high complexity of military battlefield scene,it is extremely urgent to improve the detection capability of infrared small target detection technology under complex background.Even though the features of small infrared targets have been widely mined,the suppression effect of strong background clutter and the detection effect of small infrared targets with low signal-to-clutter ratio still need to be improved,and the mining of detailed features of small infrared targets is becoming more and more difficult.This paper mainly studies the following contents to explore the infrared small target detection method under complex background and the low signal-to-clutter ratio infrared small target detection method.Firstly,the characteristics of infrared image and infrared small target are studied.In order to further distinguish the more detailed difference between the infrared small target and the complex background and to discover the characteristics of the infrared small target,the DotCurve system including various characteristic parameters of the target and background was innovatively proposed,the characteristics of small infrared targets and complex backgrounds under the system are analyzed.In addition,the imaging model,image noise and non-uniformity are analyzed,which lays the foundation for target detection.Then,for the detection of small infrared targets in complex backgrounds,the Dot-Curve system is used to analyze the detection results of small infrared target detection algorithms based on curvature.The reasons for the false alarms were explored,and the different characteristics of the complex background and small infrared targets under the Dot-Curve system were found.The similarity of target is used to quantify the difference and the curvature detection results are corrected.An infrared small target detection algorithm based on curvature and the similarity of target is proposed,which solves the problem of detecting small infrared targets in a complex background.In terms of signal-to-clutter ratio and background suppression factor,compared with the infrared small target detection algorithm based on curvature,the average increase is 2.21 times and 1.5 times,and the false alarm rate is reduced under the same detection rate.For the detected results,a tracking algorithm based on multi-feature cosine similarity is proposed to further eliminate false targets and retain real targets.Next,for the detection of small infrared targets with low signal-to-clutter ratio,according to the characteristics of low-signal-to-clutter ratio targets in the Dot-Curve system,the infrared small target detection algorithm based on curvature and the similarity of target is optimized to realize the detection of small targets with a signal-to-clutter ratio of 5.In addition,the method of extracting area targets at multiple scales solves the problem that area targets are difficult to detect.An image fusion method based on target detection is proposed,which enhances the targets' signal-to-clutter ratio while protecting the image details of the fusion image.Finally,the proposed detection algorithm for small infrared targets in the complex background and the image fusion algorithm based on target detection are transplanted to the ZYNQ platform.Experiments in the field verify the practicability of the system.