期刊論文
| 學年 | 89 |
|---|---|
| 學期 | 1 |
| 出版（發表）日期 | 2001-01-01 |
| 作品名稱 | An intelligent content-based image retrieval system based on color, shape, and spatial relations |
| 作品名稱（其他語言） | |
| 著者 | Shih, Timothy K. |
| 單位 | 淡江大學資訊工程學系 |
| 出版者 | 臺北市：行政院國家科學委員會 |
| 著錄名稱、卷期、頁數 | Proceedings of the National Science Council: Part A, Physical Science and Engineering=行政院國家科學委員會研究彙刊：自然科學與工程 25(4), pp.232-243 |
| 摘要 | Content-based multimedia information retrieval is an interesting but difficult area of research. Current approaches include the use of color, texture, and shape information. In this paper, we present a hybrid approach, which incorporates color, shape and spatial relations among objects in a picture, and has been used to retrieve images from nearly 5000 pictures. A revised color scheme and its indexing technique are used to improve the efficiency of retrieval, based on our clustering method and color sensation. A seed filling-like mechanism is used to extract shape and spatial relations from objects. A qualitative approach is applied to similarity comparison of spatial differences. The system is also implemented with a friendly GUI, which allows sketch images as well as relevance feedback to improve the accuracy of retrieval. Our experience shows that the system is able to retrieve image information efficiently based on the proposed approach. Content-based multimedia information retrieval is an interesting but difficult area of research. Current approaches include the use of color, texture, and shape information. In this paper, we present a hybrid approach, which incorporates color, shape and spatial relations among objects in a picture, and has been used to retrieve images from nearly 5000 pictures. A revised color scheme and its indexing technique are used to improve the efficiency of retrieval, based on our clustering method and color sensation. A seed filling-like mechanism is used to extract shape and spatial relations from objects. A qualitative approach is applied to similarity comparison of spatial differences. The system is also implemented with a friendly GUI, which allows sketch images as well as relevance feedback to improve the accuracy of retrieval. Our experience shows that the system is able to retrieve image information efficiently based on the proposed approach. 內容式多媒體資訊擷取是一既有趣但又困難的研究領域，現今所用之主要方法包括顏色（color）、紋理（texture）、形狀（shape）等特徵。本論文提出了一包含顏色、形狀及影像中物件的空間關係之方法，以便從近5000張的圖片中查詢所需之影像。另外，根據自訂的顏色叢集方法及人類對顏色的知覺，我們提出一套自動產生影像資料庫索引的機制，此索引機制可使影像的過濾更有效率。而且，我們提出了一個類似播種（Seed-Filling）的演算法，可以順利的求得影像中物件的形狀及空間的關係。由於一般而言，使用者很難確定物件間的距離，所以我們使用定性的（qualitative）空間關係來分析空間的相似性。另外，本系統所提供的圖形式使用者介面及繪圖工具，可以讓使用者以載入或描繪影像的方式，輕鬆的查詢影像。除此之外，本系統的回饋學習機制更可增強搜尋的準確度。我們的經驗證明本論文所提之方法，可讓使用者有效地找出所求之影像。 |
| 關鍵字 | content-based image retrieval; image database; color; shape; spatial relation |
| 語言 | en |
| ISSN | 0255-6588 |
| 期刊性質 | 國內 |
| 收錄於 | |
| 產學合作 | |
| 通訊作者 | |
| 審稿制度 | |
| 國別 | TWN |
| 公開徵稿 | |
| 出版型式 | 紙本 |
| 相關連結 |
機構典藏連結 ( http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/59869 ) |