Sea Surface Ship Target Fusion Detection Method Of Remote Sensing Image

With the rapid development of remote sensing technology,more and more attention has been paid to ship target detection technology based on remote sensing images for its widely use in civilian and military fields.In this paper,the optical imaging remote sensing satellites with different orbits and different resolutions are used to study the ship target detection and the dynamic tracking of multi ship targets in large area.The complementary advantages of regional continuous imaging of geostationary orbit remote sensing satellites and high spatial resolution imaging of low-orbit remote sensing satellites are realized,and the efficiency of target detection and tracking is improved.The main work of this paper is:1.In the high spatial resolution satellite images,the visual saliency model is introduced into the ship target detection to overcome the shortcomings of the traditional target detection method,whose analysis is difficult and the calculation efficiency is low.The effects of different visual saliency models on ship target detection is analyzed and compared.The experimental result shows that the visual saliency model based on histogram contrast is superior to other models in both detection effect and operation time.By using this model combined with space dimensionality reduction algorithm to detect the ship targets,the experimental result shows that this method not only reduces the time consumed by the target detection to the original 10%~12%,but also weakens the influence of complex surface texture background on target detection.2.Several typical background suppression algorithms are studied,and the top-hat morphological filtering algorithm is improved.The method of small target detection based on geostationary orbit satellite image is studied,and the fusion detection and discrimination method of ship target based on multi-source satellite images is designed.This method not only overcome the disadvantage of low reliability of the geostationary orbit remote sensing satellite,but also overcome the weak timeliness of low orbit high resolution remote sensing satellites,so it can continuously monitor ship targets in a wide range of sea areas.3.The basic principle of the particle filter algorithm is studied,and the target tracking problem based on the geostationary orbit optical imaging satellite observation is modeled and simulated.Based on the simulation of the geostationary orbit satellite sequence remote sensing image,the particle filter algorithm is used to solve the tracking problem of single small target.The tracking algorithm of multiple small targets is studied.The nearest neighbor method is used to correlate the target data,and then the particle filter algorithm is used to track the related targets.The effectiveness of the algorithm is proved by simulation experiment.The main study achievements of this paper: 1)The rapid detection method of sea surface ship target based on single high spatial resolution optical satellite remote sensing image is designed;2)A ship target fusion detection and discrimination method based on satellite images with different orbits and different resolutions is proposed;3)A tracking algorithm of multiple small targets based on nearest neighbor method and particle filter algorithm is proposed.