Technology Of Image Content Analysis And Application In Education

With the fast development of modern information technology ( IT ), education informationization has become an important part of Chinese Educational reforms and developments nowadays. The construction of education resources as an important job has aroused people's wide attention and the rational and effective use of education resources is playing an important role in enhancing education quality and promoting education informationization development. As an important information carrier, images are essential Web information resources as well as an important part of multimedia education resources. Compared with the blasting increase of information, information retrieval seems to develop much slowly, so does the image retrieval. How to extract useful image resources from plenty of information becomes a problem that educational workers care for.Facing rich Web-oriented elementary education image resources, we firstly download image data and the web page that image embedded in with multi-media search engine and web page analysis program, analyze texts around the image in web page, extract the name and key words describing the visual characteristic of the main image object and build the image index database. Then we use the technology of image analysis and image division based on the edge to extract the main object and to recognize (classify) the visual characteristic. In the end, we utilize the main object visual characteristic repository to recognize the main image object, and then to write the recognition results and visual characteristic back to the image index database, and finally to carry out content automatic index of Web elementary education image resources combining with.The paper emphasizes methods of image dominant hue extraction, shape classification recognition, texture classification recognition, image division and extraction and recognition of main image object. We make use of improved HS * B color space and extremely large value of histogram method to extract image dominant hue, minor wave alternation and Gabor wave filtering to extract texture characteristic, invariable regulation parameter to describe shape characteristic, BP neural networks and Support Vector Mechanism to classify and recognize the texture and the shape characteristic, the edge detecting and mathematics morphology method to divide an image, area to extract the main object of the image , "and or " knowledge expression to describe the main object's visual characteristic and then to build the main object visual characteristic knowledge database.The paper uses the program development tool of Visual C++ 6.0 and Matlab 6.5 to develop the image content analysis and main object automatic identification system, based on Windows XP OS and Microsoft Access 2003 database. The system downloaded 120,000 images related to subject terms in elementary education elementary school textbooks (Including Chinese, mathematics, society, science, ideology, life and so on ), carried out the experiment of image content analysis and extraction of the main object of image and obtained relatively satisfied result.The research indicates that it will enhance the image recognition accuracy rate greatly and improve index depth effectively to make full use of the main image object semantic information extracted from the Web in image content analysis and main object recognition based on the repository. This improves search accuracy rate of elementary education image resources and thereby provides the forceful guarantee to make an effective use of these image resources.