| Infrared imaging technology was first used in the military field for precise guidance,search and tracking,etc.due to its advantages of strong concealment,strong adaptability to different environments,and strong anti-interference ability.It has also been greatly expanded.In the process of actually acquiring images,due to the influence of complex environmental conditions,long-distance imaging,and high-speed variability of targets and backgrounds,the acquired images will show the characteristics of overall blur,lack of details,and inconspicuous contrast.The detection difficulty of small objects.Therefore,the improvement of infrared small target detection technology is still important.Although scholars have done a lot of related research in recent years,with the continuous improvement of technical requirements,the existing algorithms still need to be improved.Therefore,this paper discusses the existing problems,and the main research contents are as follows:(1)In order to detect small infrared targets with high detection rate,low false alarm rate and high detection speed,this paper proposes a small infrared target detection algorithm based on the combination of curvature filtering and visual saliency.The algorithm first uses curvature filtering to estimate the background in the image,and then makes a difference on the estimated results,thereby significantly suppressing most of the background information in the image;then,uses the morphological method to weaken the residual high-intensity clutter,and through the local Contrast algorithm further performs clutter suppression and target enhancement;finally,threshold segmentation is performed to obtain significant small targets.Experimental comparison shows that the algorithm has higher detection accuracy and lower false alarm rate.(2)Aiming at the problem that existing algorithms are susceptible to the interference of target type and clutter shape,which leads to unsatisfactory detection results under low signal-to-noise ratio conditions,a small infrared target detection method based on density peak search and local characteristic discrimination is proposed.Firstly,the high-frequency components in the image are extracted and fused with the original image to obtain the target enhanced image;then the candidate target points are extracted by using the improved density peak search method;then,a new method is proposed combining the local gray level difference and gradient isotropy of the target.A local discriminant model is used to locate the target area,and the curvature filter method is used to suppress the background,and the high-frequency component image is fused to enhance the target.Finally,the threshold segmentation is performed to obtain the result map.Experimental results show that the algorithm shows significant advantages in different scenarios. |
