Research On Detection Algorithms For Infrared Small Target On Sea Background

As countries around the world pay more and more attention to the marine field,various studies based on the sea background have become particularly important.As an important branch of modern detection technology,the use of image processing means to achieve the detection and tracking of sea surface targets can more quickly and intelligently obtain the relevant information of the target.Because the sea background is time-varying and random,more advanced detection methods are needed to improve the speed and accuracy of target detection.This thesis mainly aims at detection and tracking of small target under the sea background using image processing schemes.The main research contents of this article are as follows:First of all,the measured infrared data based on the sea background is acquired by the thermal imaging gimbal camera equipped on a drone,and the measured data is analyzed from the perspective of infrared imaging characteristics.According to the characteristics of the measured infrared image,a background segmentation method based on sky-sea line is proposed to realize the extraction of the Region Of Interest(ROI)of the original infrared image.Not only can it reduce irrelevant background interference,but also it greatly reduces the time-consumptions of the corresponding subsequent detection algorithms.Then,by analyzing the difference on features between the infrared small target and the sea background,a infrared small target detection algorithm named weighted patch-based contrast measure(WPCM)is proposed,which uses patch contrast and local entropy to achieve small target detection,and then get the position of the target using visual transfer mechanism.On the basis of the detection results,the contour of the small target is compensated,and the proportional feature of its circumscribed rectangle is calculated.The proportional feature is used to roughly classify the common ship targets and buoy targets in the sea background,which enhances the practicality of the algorithm.The simulation results show that the algorithm proposed in this paper has a good suppression effect on the interference of offshore clutter,ship wakes and pixel-sized noises with high brightness(PNHB),and it can also obtain higher signal-to-noise ratio gain in the face of different sea environments.Finally,for the feature that the single frame detection algorithm does not use the motion information between the target frames and is not robust enough to handle complex scenes,a feature fusion based on infrared saliency features and speeded up robust features(SURF)is proposed particle filter tracking algorithm.The corresponding feature observation model is established by multi-feature fusion.The weight of the particles is calculated by the similarity difference between the target template and the candidate template.Finally,the target state is estimated by the position information of the particles and the size of the weight.The algorithm proposed in this thesis has a stable tracking performance on the moving ship target in the measured data set.