Intelligent Processing And Recognition Method Of Large And Medium-sized Target Information On The Sea

The detection of large and medium-sized targets at sea has always restricted the maritime confrontation environment of various countries.Therefore,it is of great significance to study large and medium-sized targets at sea,such as ships,and enhance the detection ability of targets,for the maintenance of national security.There are many kinds of detection methods for large and medium-sized targets on the sea,among which optical imaging and radar imaging are two important means,which have different advantages.Therefore,how to make full use of the characteristics of optical image and radar image at the same time to improve the recognition rate of large and medium targets on the sea is worth considering;In addition,in the sea detection,the images of the same large and medium-sized target from different perspectives can express more information about the target.How to use the information from different perspectives to further improve the recognition rate of the target can also be considered.Starting from the above-mentioned directions worthy of consideration,the intelligent processing and recognition methods of large and medium-sized target information on the sea are studied in this paper.Firstly,taking five kinds of ship targets as special cases of large and medium-sized targets on the sea,the ISAR imaging and optical imaging of large and medium-sized targets on the sea are simulated,and the simulation image data sets of five kinds of targets at different angles are obtained;Considering the fact that there may be less optical images of large and medium-sized targets on the sea in practical application,the method stylegan2 based on small sample amplification is proposed to amplify the target image,and the amplified image data set is obtained,thus the overall construction of the data set is completed.Secondly,the optical image of large and medium targets on the sea is processed by using intelligent processing technology.Optical image acquisition is easily affected by external conditions,such as meteorological conditions,other target occlusion,etc.,so the optical image often has the problem of incomplete target information,which affects the further target detection and recognition.In order to improve the visual effect and overall quality of the optical image and realize the target detection and recognition more effectively,the typical problems that may exist in the optical image are solved based on pix2pix network.In order to improve the recognition rate of large and medium-sized targets at sea,this paper focuses on the relevant algorithms of the conversion and fusion of radar images to optical images of large and medium-sized targets at sea.Intelligent processing methods include four kinds of network used for image conversion,namely,cyclegan network,cut network,dclgan network and u-gat-it.The performance of various algorithms is evaluated by quantitative and qualitative means,the advantages and disadvantages of various algorithms are analyzed,and the conversion and fusion of radar images and optical images of large and medium-sized targets at sea are realized.The image generated by the above intelligent means is obtained,which improves the recognition rate of large and medium-sized targets at sea.Finally,this paper studies the problem of cross view conversion and fusion of large and medium-sized target images on the sea.In the actual sea observation,with the movement of the detection platform,the images of large and medium-sized targets at sea from different perspectives can be obtained.Based on the information of the same target from different perspectives,the cross view conversion and fusion from radar image to optical image is preliminarily completed.In general,the information of large and medium-sized targets on the sea is processed by intelligent technology,which effectively improves the recognition rate of large and medium-sized targets on the sea.