Research Report of multimedia information retrieval _ Search engine

Research Report on multimedia information retrieval <?xml:namespace prefix = o ns = "Urn:schemas-microsoft-com:office:office"/>

1. Introduction

The internet has been developed around the world, with a huge influx of information, to find what they want on the internet has become increasingly difficult, in order to meet the needs of people, many companies have launched their own internet-based search services, such as Google,yahoo, Bai Du, 3721 and so on search engines, which to a large extent, to provide people with access to information convenience. But this does not solve the problem completely, because now almost all search engines provide the function is based on ordinary text search, the various formats of documents, spreadsheet search is very little, and the multimedia information retrieval is almost lost. So there is an urgent need for a means to retrieve huge multimedia information.

2. Multimedia

As information transmission, entertainment and many other uses, multimedia has developed a relatively large and perfect. Today's multimedia information is mainly divided into 3 categories: video, audio and image. And each class is divided into many small categories, such as video includes many formats of video information: RealMedia, DivX, images are divided into JPG, GIF and other images. And these large classes are also cross places, some of the media contains both video information and audio information, and even image information. Because there are so many kinds of multimedia, it is very difficult to search for all the multimedia information, which is the direction of our research.

3. Research on the pretext of multimedia information retrieval

The multimedia information retrieval mainly aims at the multimedia information content retrieval, for instance the audio content, the image content and so on. Because different to the media has certain difference, therefore classifies the research to search the excuse. Now assume that all multimedia information is available, that is, there is no need to specifically search for those multimedia information.

A) video. We usually see a variety of video information has a variety of film files, such as various RMVB, ASF, MPEG, AVI and so on. There are a variety of ways to retrieve these videos, such as using the name of the movie to retrieve, using the words of the movie's mission to retrieve, using the film to retrieve each specific lens, and even use a series of footage to retrieve and use a storyline to retrieve. Of course, you can also use sound to retrieve, but this is the range of audio retrieval.

b) Audio. Audio contains a variety of music files, such as MP3, WAV, and even most video files contain audio. If there is, you can use the lyrics name, singer name, song name to retrieve. You can also use music to retrieve, such as humming a tune, and then you can use this melody to retrieve. can also use the voice of the intensity of change, voice expressed a feeling of emotional atmosphere, such as retrieval.

c) images. Now the image file is already a variety of, it can express the rich and colorful information. Can be used to retrieve the characteristics of the image itself, such as the texture of the image, color, shading to retrieve. You can also use a thumbnail to retrieve, such as a sketch, and you can search for images that are similar to a draft. can also use the information expressed by the image to retrieve, such as using the word "man" can search all the pictures of men, with the word "Jordan" can find a picture of Jordan.

4, research on the realization of information retrieval function

Multimedia is different from ordinary text, it expresses the content of complexity, abstraction is much higher than ordinary text, so the difficulty of its retrieval is far more than ordinary text. It was proposed that: can be a different multimedia information manually processing, where needed to add comments, the retrieval of information only need to retrieve comments on it, which is feasible in a small amount of information, if the information rapidly increase, then the annotation of multimedia information costs of human resources is unacceptable. It has also been suggested that all kinds of media information can be converted into ordinary text, and then indexed structure can be indexed query, this scheme can be used to reuse the existing text search technology, but how to convert multimedia information into ordinary text is a big problem. Another is to redesign the format of the media information standards, so that this format can be fully expressive and convenient to be queried, this is also a research direction, the main problem is the formulation of this format standard.

5. Summary

With the development of multimedia technology, may be more and more kinds of multimedia, may not stop the above, but the overall direction should not be too far. The main problem now is to have ideas, but there is no technical capacity to achieve them.