| Summary: | 碩士 === 國立臺南大學 === 資訊工程學系碩士班 === 101 === Digital photos and images increase considerably in recent years. How to manage and access images efficiently and effectively becomes an important issue. The main classes of image retrieval approaches contain the content-based retrieval approach and the keyword-based retrieval approach. The content-based retrieval approach searches images with a similar query image, but getting a suitable image is also a problem. In other hand, the keyword-based retrieval approach searches images with keywords. However, annotating images manually by users is not a good solution. Thus, the requests of automatically image annotation increase with increasing images. There are many previous researches on image annotation, but the annotation results are still not good enough to be used. Therefore, we propose a hybrid framework for automatically image annotation in this thesis. The framework combines the previous researches and considers the relationship between each keyword. The proposed framework can annotate images efficiently.
|
|---|
