Image Recognition Of Ship Targets Based On Cloud Computing

Information warfare will be the main form of combat in the future war, the warring parties have attempted to get control of information by controlling the battlefields' intelligence and information. Much investigation equipment including satellite have become an important source of access to information. Satellites' reconnaissance which has a wide range, high speed, under the threat of small advantages has been widely used in many aspects. As an important investigation, Satellite imagery satellite data have a strong practical significance and national defense significance.Our country inland sea and side of the sea's area has about 470 square kilometers. As an important means of transport and combat strength in the sea, the warships in the maintenance of Marine safety and good Marine order play an important role.In this paper, the main work is as follows:(1) We solved the problem which lack of the satellites images of the warships target when we study it. By using the Google Earth Pro to collect the world's largest port, we have many warships target images of the same area in different period.(2) A system framework model is established for the acquisition of the satellite image, the target location of the target ship and the target recognition. And the common methods of image recognition and classification are described, which provides reference for the realization of the latter system.(3) We realize to get suspicious target on Baidu map platform, through processing the difference in different period of satellite image, and put forward the solution to solve the suspicious target's longitude and latitude according to the original longitude and latitude of satellite images. And we has carried on the practice and application in the Baidu map.(4) Image recognition efficiency issues in the warship target recognition is solved. Three methods, scale-invariant feature transform, histogram of oriented gradients and deep learning, are adopted in the identification process of the two-type warship recognition. Because of the high accuracy of deep learning used in the two-type recognition process, the three-type and four-type issues are deeply studied by the same method.(5) We build OpenStack cloud platform in the realization of Digits system. And the system is completed in the environment of the network structure to GoogLeNet depth based on two target ship classification, and get good results.