The Research And Implementation Of Image Retrieval Algorithms Based On Compressive Sensing

Recently,with the rapid development of the Internet and multimedia technology,people's life is more and more closely connected with digital images.This phenomenon is based on the popularity of the various digital equipment.However,the effective methods solving these amounts of image data has been very necessary.Image retrieval study started in the 1970s.Until the 1990s the content-based image retrieval method become a milestone of the study.The key of the content-based image retrieval technology is extracting the image features.The features should represent the image information extremely.Features extraction is also the hot spot problem.In 2004,a novel sampling theory(compressive sensing)was born.It broke through the bondage of traditional sampling theory and brought a revolutionary breakthrough in signal acquisition field.As a result,the image retrieval technology based on compressive theory rises in response to the proper time and conditions.It opens the door to the new research field.It can raise the query accuracy.What's more,it's able to reduce the search time.Therefore,the study of image retrieval based on compressive sensing has important practical significance and prospect.This thesis started with the basic theory of compressive sensing,and it analyzes this technology's research status and development.Meanwhile it introduces the design principle of the content-based image retrieval technology.Consequently,this thesis proposes two kinds of novel image retrieval algorithm based on compressive sensing,which are the block image retrieval scheme and the compressed domain image retrieval scheme.It presents these method's design principle and implementation in detail.And conducting the contrast experiments with the classic content-based image retrieval method.The experimental results show that the proposed two kinds of image retrieval algorithms have high precision.At the same time,the method owns the high precision of recall comparing with the traditional content-based image retrieval.Simultaneously,these methods save the retrieval time and improve the retrieval efficiency.The new methods are the remarkable innovation.The image retrieval method based on compressive sensing has the advantage of high retrieval accuracy and speed.Nevertheless,this kind of retrieval method is in the exploration stage.The block image retrieval method based on compressive sensing is suitable for precise image retrieval.And the compressed domain image retrieval method is able to retrieve compressed images directly,such as searching smapling remote sensing images without recovery.