Research On Infrared Small Target Detection Algorithm Under Complex Background

With the continuous development of precision guided weapons,long-range precision strike technology has gradually become the dominant factor affecting the process of war.Correspondingly,air defense early warning systems have received more and more attention from military industry departments in various countries.In the air defense early warning system,infrared imaging technology occupies an important position by virtue of its strong concealment,long detection range,all-weather,anti-electromagnetic interference,and strong mobility.However,in actual application scenarios,due to the long detection distance,the target in the infrared image usually has problems such as small size,weak brightness,lack of structure and texture information,and the imaging will be affected by atmospheric attenuation,bad weather and noise.Therefore,Small infrared targets are easily submerged in complex background clutter.This article first builds a mathematical model of infrared images,focusing on analyzing the characteristics of small targets,background and noise in infrared images,providing a theoretical basis for subsequent research.In order to improve the detection performance of the infrared early warning system,this paper carried out an in-depth study on the infrared small target detection algorithm under complex background.The main research contents are as follows:Aiming at the problem that the existing infrared small target detection algorithm has an unsatisfactory suppression effect on complex backgrounds such as sky,clouds,sea and sky,this paper deeply analyzes the grayscale and gradient features of small infrared targets and background areas,and proposes an infrared small target detection algorithm that combines grayscale features and gradient features.First of all,this paper uses the feature that the gray value of the small target is higher than the neighborhood background to enhance the target intensity and suppress the uniform background.Secondly,considering that the gradient directions of small targets are different and generally point to the center,and the gradient directions of sharp edge backgrounds have a high degree of consistency,the gradient characteristics of small targets and backgrounds are used to further suppress the edge background.Finally,an adaptive threshold algorithm is used to segment the target.Experiments show that the algorithm can significantly increase the intensity of small targets under different complex backgrounds,effectively suppress complex background clutter,and has good detection performance and robustness.Traditional detection algorithms use artificial design features for target enhancement and background suppression,which obviously lacks robustness in complex scenes with strong radiation sources such as surface vegetation and buildings.In order to make up for the lack of features of infrared images and the defects that the features are difficult to describe,this paper proposes an infrared small target detection algorithm based on region selection and convolutional neural network,using the feature extraction ability of convolutional neural network to fully explore the potential features of infrared images.This paper draws on the idea of a two-stage target detection algorithm.In the first stage,the grayscale characteristics of the infrared image are used for preprocessing,and then the candidate area is extracted by the corner detection method.Considering that there may be confusing background clutter in the candidate area,the convolutional neural network classification model is used to classify the candidate area in the second stage,and non-target areas are filtered out.Finally,a large number of comparative experiments have proved that the algorithm can greatly reduce the false alarm rate while ensuring the detection rate.