Research On Detection Technology Of Weak And Small Target In Strong Light Background

Affected by factors such as illumination,the visible light detection system suffers from serious non-uniformity during imaging,so that weak and small targets are submerged by strong clutter,which makes it difficult to detect long-distance space targets.In this paper,on the one hand,the preprocessing method is used to correct the image non-uniformity to enhance the target LSNR and contrast.On the other hand,a weak and small target detection method based on gradient feature extraction is proposed for high background intensity,low LSNR and weak contrast of visible images under strong light.The main research works are as follows:(1)Target enhancement: a non-uniformity correction preprocessing method based on local power transformation is proposed to improve target SNR and target enhancement effect.Under the influence of strong vignetting non-uniformity,the correlation between weak targets and the background becomes stronger,which limits the effect of background suppression and target enhancement,making it difficult for traditional algorithms to detect.Vignetting non-uniformity in strong light background images is widely distributed and irregular,and it is difficult to completely eliminate by modeling.In addition,the non-uniformity correction method of natural scenes usually corrects the global image,which is not conducive to protecting the target.Therefore,this paper uses different γ values to perform power transformation on local areas of different gray levels to obtain a stable output image,and at the same time uses the stretching effect of power transformation to enhance the contrast between the target and the local background.The algorithm does not require mathematical modeling of vignetting non-uniformity and can be used as a preprocessing stage before enhancement and detection.The experimental results show that the non-uniformity correction method of the local power transform can improve the LSNR of the traditional background suppression and target enhancement algorithms by more than 1.05 d B,and can fully detect the sequence images with less background clutter and noise.(2)Weak and small target detection algorithm: Aiming at the problem that traditional methods are difficult to detect small targets with low contrast and low LSNR,a weak target detection method based on gradient feature extraction is proposed.Traditional single-frame detection methods usually rely on the discontinuity of target and background and do not make full use of the shape characteristics of targets,resulting in low LSNR and low contrast target low detection rate and high false alarm rate under the influence of strong clutter.In this paper,using the shape of weak and small targets and the distribution characteristics of isotropy in the gradient space,a gradient feature extraction template is designed to enhance the target while suppressing the background,and then the target gradient feature is enhanced by local contrast to further improve the target SNR.Finally,target detection is achieved by threshold segmentation.The experimental results show that the weak target detection algorithm based on gradient feature extraction proposed in this paper can directly detect weak and small targets with a strong light background of about-2d B,and can accurately detect a strong light background of about-3d B after combining with the non-uniformity preprocessing algorithm.