Research On Detection Methods Of The Infrared Dim Small Target In The Complex Background

The detection technology of infrared dim and small targets in complex background is of great significance for the development of infrared early warning,precision guidance and other fields.The earlier the target is detected,the more reaction time will be provided for the back-end,so it is very important to achieve effective detection in a long distance.Compared with the traditional radar system,the infrared search and tracking system is a passive working system.It can identify the target by receiving different radiation energy of the object,and it has the advantages of small size,light weight and so on.It is suitable for installation in fighter or drone.Small and weak targets occupy very few pixels in the image,and there are no details and texture features of the object.In the imaging process,affected by background clouds,atmospheric radiation and other factors,the received target intensity is greatly reduced,which often causes the target radiation intensity lower than the non-target area in the infrared image,resulting in the phenomenon of target loss and high false alarm probability when detecting the target.In addition,infrared dim small target detection technology is an important component of infrared search and tracking system,and is the key technology of military,aviation and aerospace applications,so the detection of infrared dim small target in complex background is a very practical research direction.In this paper,starting from the imaging characteristics of infrared small and weak targets,the methods of background suppression,threshold segmentation and joint detection in space-time domain are studied,and the detection methods that can effectively solve the dispersion and multi morphological characteristics of small and weak targets are proposed,and finally the effective detection of infrared small and weak targets in complex background is realized.Firstly,the imaging characteristics of infrared dim small target are studied in detail.Because of the long detection distance,the imaging area of the target is small,and the background is mostly complex background such as large area clouds or buildings,so the image must be preprocessed first.In the infrared target detection of single frame image,in order to detect the target with low SNR,a new background suppression method is proposed to solve the problems of complex image background and diversity of target shape.In the preprocessing part,considering the diffusion of the target and the dynamic distribution characteristics of the target in the image plane,we improve on the basis of soft morphology.Combined with the multi-directional dynamic filter template,the filter template selects 5 × 5 structural elements in four directions,which can effectively suppress the background,retain the shape information of the target to the greatest extent,and have better robustness.In the part of threshold segmentation,a multi-directional gradient threshold segmentation method is proposed,which counts the gradient values of 8 directions that are not synchronized,and counts the number of gradients that exceed the threshold to filter the target points.Finally,the remaining false alarm noise can be eliminated by energy concentration.Experiments show that the proposed detection method can effectively improve the target detection rate and reduce the false alarm rate,the detection rate reaches 98.5%.Compared with other algorithms,the algorithm proposed in this paper improves the detection rate by 20% and 14%,and reduces the false alarm rate by 52.6% and 44.4%,respectively.Single frame detection is easy to cause the problem of missed detection and high false alarm rate.In this paper,background modeling and data association are combined to propose a new joint detection algorithm in space-time domain.Firstly,the gray value of each pixel in the image is related to the gray value of neighboring pixels.According to this feature,a model is established to estimate the background,and most of the background is removed by difference with the original image.Then,the remaining false alarm points can be eliminated and the detection rate can be improved by using the energy concentration criterion.According to the continuity of moving target in multi frame image,particle filter is used to track the target track and determine the target position.Experiments show that the proposed algorithm can effectively track the target track and achieve multi frame target detection.