Research On Ship Recognition Based On High Resolution Remote Sensing Image

The 21st Century is the ocean age.In order to protect their maritime rights and interests,all countries have intensified their supervision over the sea area.Every country is focused on the supervision of ships in their own seas.Ship monitoring is related to the development of national economy and national security.The ship monitoring not only can be used for shipwreck rescue work,but also can be applied to combat illegal fishing vessels,illegal dumping vessels,smuggling ship and anti piracy and other acts.With the rapid development of human marine activities,the research on ship target monitoring has been paid more and more attention.At present,Chinese main ship dynamic monitoring system includes VTS(vessel traffic service),AIS(automatic identification system),video surveillance system and sea patrol system.However,the dynamic monitoring system cannot meet the needs of maritime dynamic supervision.In this paper,the application of high resolution remote sensing image to ship target detection and classification is taken as its research direction,aiming at using new remote sensing technology to detect the ships.The high resolution remote sensing image will be a supplementary means for dynamic monitoring of ships.In order to meet the needs of the ship dynamic monitoring system,this paper designs a set of high resolution optical satellite remote sensing imagery for ship target fast detection and classification.Firstly,one dimension Otsu algorithm is improved.The Otsu algorithm based on principal component analysis is proposed to improve the efficiency and noise resistance of the extraction of the optical remote sensing image.After that,we analyze different types of ship texture by using gray level co-occurrence matrix.By this,we model different types of ships.Finally,the fuzzy pattern recognition algorithm is used to classify the ship based on the texture features of the ship.In this paper,the experiment of ship target detection and classification is carried out on GF-1 optical remote sensing image.From the experimental results,we find that the Otsu algorithm based on principal component analysis has greatly improved the accuracy and efficiency,while the accuracy rate of ship classification results is over 90%.The ship fast recognition process designed in this paper has good stability and it can be used as a powerful complement to the ship dynamic identification system.Classification method of ship detection is designed to quickly ship target recognition in large area.The method designed in this paper has a high accuracy and practical application value in the field of rapid detection of remote ship.It will provide new technical methods for technical supervision of ship dynamic in china.