A SURVEY ON CONTENT BASED VIDEO RETRIEVAL USING MPEG-7 VISUAL DESCRIPTORS
International Journal of Advance Engineering and Research Development
The volume of the video content grows very fast and most of the video search systems are based on manual annotations or use text information. But such information is not always of high quality or lack precision. In the field of multimedia technology, video retrieval has become one of the fastest growing research areas. Several methods have been developed for retrieval of videos based on extracting their visual features automatically in recent years. The most commonly used low level visual
... es are colour, texture, shape, motion and spatial-temporal composition. The use of visual descriptors of MPEG 7 is to provide interoperability of system because of its standardization. This paper describes the applications, challenges, methods and limitations of content based video retrieval systems. It also explains the MPEG-7 visual descriptors with their application. Keywords-Content based video retrieval, MPEG-7, shot detection, key frame extraction, visual descriptor. I. INTRODUCTION Content-based video retrieval (CBVR)    is the solution to video retrieval problem which in turn is the problem of searching for digital videos in large databases. The meaning of "content-based" to search and analyzes the contents of the video rather than the textual metadata or descriptions associated with the video. The term "content" refers to the low level information such as colours, shapes, textures, or any other can be derived from the video itself. CBVR is desirable because earlier multimedia searches purely rely on metadata which in turn dependent on annotation quality and completeness. When humans manually annotate videos     by entering keywords or metadata in a large database, it can become time consuming and may not capture the keywords desired to describe the video. The evaluation of the effectiveness of keyword video search is subjective and has not been well-defined. The large amount of the multimedia content information generates a great need for efficient techniques of finding, accessing, filtering and managing multimedia data. Before retrieving the video based on its content, some pre-processing has to be performed on video like video segmentation, key frame extraction and feature extraction. Multimedia database management systems use the query-by example paradigm to respond to user queries. Users are needed to formulate their queries by providing examples. One of the important issues to be considered in today's multimedia systems is interoperability: the ability of diverse systems and organizations to work together (interoperate) . This is very critical for distributed architectures if the system is to be used by multiple clients. Therefore, MPEG-7 standard as the multimedia content description interface can be employed to address this issue. MPEG-7 is a multimedia content description standard. It was standardized in ISO/IEC 15938 (Multimedia content description interface) . This description will be associated with the content itself, to allow fast and efficient searching for material that is of interest to the user. MPEG-7 is not a standard for encoding of moving pictures and audio. MPEG-7, formally named "Multimedia Content Description Interface," is the standard that describes multimedia content so users can search, browse, and retrieve that content more efficiently and effectively than today's mainly textually annotated search engines. It's a standard for describing the features of multimedia content. CBVR is useful in many applications like news broadcasting, advertising, searching music video clips, distant learning, video archiving, medical applications etc. CBVR systems provide the efficient and more accurate way to retrieve the videos. But it also faces some challenges listed below: Very large collection of video data still requires significant time to compute the features. Semantic information retrieval is also a major issue in CBVR. It is very challenging task to choose the feature that reflects the real human interest. It is difficult for CBVR systems to support multimodal query and allow flexibility for the user to specify its query parameters. The CBVR should also provide different search strategies adapted to the type of search to the user. The CBVR systems are divided into mainly four steps as shown in following Figure 1 : shot detection, key frame extraction, feature extraction and similarity measurement.