| Summary: | 碩士 === 國立中正大學 === 資訊工程所 === 98 === With the rapid growth of Internet, there are a large number of digital images. How to retrieve the digital image in database efficiently has been an important issue. Traditionally, we tag each image with several keywords, so that users can input keywords to search image. However, this method is more suitable for users who have already understood the searched object clearly. Meanwhile, an accurate query result can be obtained only with adequate keywords information. Generally, non-professional consumers search objects with an ambiguous concept. That is, probably, the name of the product is unknown, or the name contains no characteristic which they describe. That makes query result give users a low search ranking.
Clothes products contain many characteristics which are difficult to describe in keywords, such as texture, shape, or the relationship between object and space. Based on this issue, we develop an image-based visual clothing retrieval system, which extracts and uses the features of clothing images to find objects that are difficult to describe by text, or without text annotations.
In the study, we address these two issues by developing (1) extract appropriate feature description from clothes image, (2) build retrieval strategies of clothes image in database. We propose a part-based retrieval system framework, which provides three visual search conditions; that is, neck, colors and the region-of-interest, to assist in finding clothes products efficiently. And, we experiment on an image database with about 1891 general-purpose images, results showing that our approach is more helpful for picking clothes product.
|
|---|
