Research of Orthodontics Image Retrieval Based on Multi-Feature
|School||Jiangsu University of Science and Technology|
|Course||Applied Computer Technology|
|Keywords||Image Retrieval Cumulative histogram Image Segmentation Moment Invariant|
The 21st century is a century of the explosion of information, such as the wave of computers and the Internet into the way people work, learn, in all spheres of life, can be said to be everywhere. A hot field of image retrieval technology is gradually developed in this situation, it is not only a combination of more than one subject knowledge, including image processing, pattern recognition, database technology, and also to promote the development of other disciplines. The early text-based image retrieval already can not meet the growing information needs. After persistent researchers delve into the content-based image retrieval technology scholars. The user simply selects the specified search image retrieval system will automatically complete the characterization of the image, extraction, identification and matching steps. In oral medical diagnosis, have a good sense of practical application, dentistry doctor can use an image retrieval system, to find some of the worst oral picture, comparative analysis, and refer these cases the diagnosis program to improve the accuracy of diagnosis and efficiency. This paper first introduces the research background and significance, as well as content-based image retrieval technology research status. Detail and content-based image retrieval technology. This article focuses on the color-based image retrieval and shape-based image retrieval. Can not get better retrieval results for the past, using only a single image of the underlying characteristics, image color features and shape features a combination of multi-feature fusion methods for image retrieval. For color feature extraction, this paper introduces a cumulative color histogram with spatial information retrieval method, thinking: Firstly, HSV color model to quantify, and then the image is divided into multiple sub-blocks, and according to the different regions The degree of importance given their different weights, and finally combined cumulative histogram to describe color characteristics. Shape feature extraction using very classical Canny operator split method are combined with the Otsu threshold segmentation, extracted seven invariant moments to describe the shape of the image features. Multi-feature fusion methods that give weight to the color features and shape features, and feature similarity matching. Finally, Orthodontic image million image library image of these experiments, several comparative analysis of experimental results obtained based on multi-feature image retrieval method for Orthodontic image good retrieval effectiveness, can be applied in the Orthodontic aided diagnosis, help doctors improve diagnostic efficiency, so as to promote the health industry, the process of building information.