| Along with the popularization of the Internet and the rapid development of multimedia technology, a lot of multimedia information are more and more available. How rapidly and accurately obtain a huge amount of multimedia information has become urgently necessary to solve the problems. The traditional retrieval system is text retrieval system, but text is very difficult to describe the rich audio and video information, and the workload of manual tagging is heavy and subjective. Therefore, the lack of effective means of retrieval system has impeded the further development of multimedia technology. So the content-based search and retrieval technology emerges as the times requirement, which is to solve the effectiveness and efficiency of multimedia information retrieval.The content-based retrieval system describes and searches multimedia content by their own features, including images of dynamic video, audio and other forms of multimedia information. Early in the content-based retrieval system, the description of multimedia information had no uniform standards. Many companies developed multimedia information retrieval systems using different descriptor. This has also led to the retrieval system is not interchangeable, and the majority of multimedia information retrieval systems are lack of particularly effective description. Therefore, in order to meet the needs of the development of multimedia technology, MPEG-7 standard emerged. It resolved the question of the standard for describing multimedia information. It is also conducive to the development of multimedia content-based retrieval.MPEG-7 standard provides a framework of standardized tools that can be used to describe and efficiently manage multimedia content. It drew on the visual descriptors based on images of video content. Visual descriptor consist of lower-level semantic features, such as: color, shape, texture and motion, it also consist of high-level semantic description. Using MPEG-7 standard to describe visual content can be more direct and effective to express image and video features, so it promote the development of research in the field of multimedia technology. In the MPEG-7 standard, feature extraction and retrieval engine are non-standard, the part of standard is description and coding. This will provide much broader study space for researchers and better application development for businesses.FPGA is a sort of programmable logic device that has been widely using. It originated the company of Xilinx in the United States. The company manufactured the world's first piece of FPGA in 1985. In the development of 20 years, FPGA hardware architecture and software development tools are being constantly improved, and become more and more perfect. It has available door from the initial 1,200 to hundreds of millions of single FPGA gate, the system frequency is up to a few hundred MHz. With the development of the semiconductor industry and the development of deep sub-micron manufacturing, the world's top manufacturers such as Xilinx,Altera, has improved integration of FPGA device to a new level. FPGA design combining with hardware description language, enhance the capabilities of hardware circuit design, greatly reduce design time and improve the efficiency of designers. FPGA design also combine with Communication, DSP and Video technology fields, their key algorithm can be designed by FPGA and can be formed to FPGA IP modules with independent intellectual property. This is special significance for integrated circuits industry and the development of digital image and video technology. FPGA is widely used in all areas of electronic design, it has great vitality. With the improvement of FPGA, it will promote the development of modern digital wave.In the first place, the paper studies the MPEG-7 standard and visual descriptor. Briefly introduce the MPEG series, an overview of the basic concepts of MPEG-7 standard, the main content, multimedia description schemes and its applications. Highlighting describe the visual descriptor of the MPEG-7 standard, including its basic structure, color descriptors, texture descriptors, shape descriptors, motion descriptors and some specific descriptor for specific application areas such as face recognition. Also discuss the relationship between the MPEG-7 standard and content-based image retrieval system.Then, the paper introduces the field programmable gate array FPGA and modular design. Briefed the EDA and programmable device, summarize FPGA from the basic concept, the major manufacturers, the basic resources, major programming techniques, the advantages and its development, and describes top-down design method of FPGA and its development process. It also introduces the modular design with its basic concept and design process.Shape is one of the most important features of image content. Using shape to describe the image characteristics is very intuitionistic. The description of shape characteristics are often based on the shape contours or regional characteristic. Fourier transform often used to describe shape, but it is usually described in the contour feature, but loses the regional information. In order to make Fourier transform description take into account the contour and region of the image shape, we present block shape descriptor based on Fourier transform in the paper. The descriptor is one of shape descriptors in the MPEG-7 standard. In the fourth chapter, we take the focus of such shape descriptor.The extraction process of the shape descriptor based on Fourier transform consist of three main stages. The first stage is to transform the two-value function of the imported image from the Cartesian coordinate system to the polar coordinate system. In the transformation process, it is necessary to pay attention to selecting the origin of the polar coordinate system. We often choose the shape centroid as the origin of polar coordinate. This will avoid the influence of the starting point of the shape, and keep translation-invariance. The second stage is to do discrete Fourier transform for image function in the polar coordinate system, and gain the initial coefficients. In the transformation process, we often overlook the phase information of the shape. This can be maintained coefficients rotation-invariance. The third stage is to normalize the initial coefficients. The step is the primary role of keeping scale-invariance of the coefficients. After the process, we gain an eigenvector. The eigenvector shall be the subject of shape descriptors.After the study of the extraction process of the shape descriptor, we can use FPGA to implement it. In the hardware implementation process, we adopt top-down design method and modular design plan. The paper is divided into three modules to implement the algorithm of the shape descriptor, including coordinate transform module, Fourier transform module and normalized module. In the paper, we describe the design of the various modules, discuss the program of each module, and give some solutions that are synthesized and simulated.By studying the arithmetic of the MPEG-7 shape descriptor and discussing the design of FPGA, the shape descriptor can be applied to content-based image retrieval system. This paper gives the important reference value for content-based retrieval and hardware implement about MPEG-7 visual descriptor. Meanwhile, it provides great reference value for the use of FPGA high-performance image processing technology and large scale system-on-chip design (SOC) in the future. |
PDF Full Text Request