Research On Scene Image Classification And Ship Recognition Based On Feature Learning

With the development of imaging technology,the spatial resolution of image continuously improve.So that images contain a large amount of information and present more detailed information such as spatial arrangement information and textural structures which are of great help in recognizing the contents of an image.However,the image has some characteristics that are high dimension,complex structure and redundant information etc.Thus,it is very difficult to deal with original image directly.Therefore,it is necessary to represent the original image accurately and effectively by learning various features and to provide a higher discriminative image feature representation for the subsequent recognition and classification task.In this paper,the feature learning of remote sensing scene image and optical ship image is studied.The main contributions of the paper are:Firstly,On the basis of a large number of literature research,we elaborated the necessity of feature learning for the image,and then introduces several common global features and local features of spatial feature learning at the field of computer vision.Secondly,in this article,a local feature representation method using patch-based MS-CLBP features and FV is proposed.The MS-CLBP operator is applied to the partitioned dense regions to extract a set of local patch descriptors,and then the Fisher kernel representation is used to encode the local descriptors into a discriminative representation of remote sensing images.The two implementations of MS-CLBP are combined into a unified framework to build a more powerful feature representation.The proposed method is evaluated on two public benchmark remote sensing image datasets and obtains superior classification performance.Thirdly,since ship's appearance affected by external factors such as lighting or weather conditions,we proposed a multiple feature learning method using three features including a global and two local features.In this multiple features learning framework,feature-level and decision-level fusion are both investigated.We use Gabor-based MS-CLBP to extract global feature to compensate for the local feature,therefore taking full advantage of the complementary nature between global and local features.We use two optical ship image datasets to verify the effectiveness of the multiple features learning method and achieve significant recognition performance.