利用概念知識與模糊理論進行空間知識之模型建構

Applying Concept Knowledge and Fuzzy Theory to Construct Spatial Knowledge

王佳祥(Chia-Hsian Wang)

指導教授 ： 賀嘉生

中原大學/工學院/資訊工程研究所/碩士(2003年)

https://doi.org/10.6840/cycu200300595

摘要

摘　要
空間知識在生活上扮演著重要的地位，而影像對於了解空間的關係有相當幫助，藉由影像能看到未曾去過或是到不了的地方，若能夠善加利用這種媒介，即可讓學習者還未出門便能夠了解各種地區的空間關係與知識。
在本文中以影像資訊擷取為出發點，探討影像中所能夠承載的空間訊息，輔以影像資訊處理技術，取得影像當中隱含了像是顏色、紋理、形狀等屬性。影像的空間知識當中的基礎元素稱之為影像物件，影像物件是從空間的資訊的抽象化後所形成的。此後，利用影像物件的特性作為為獲取空間知識的基礎，探討人類腦中記憶空間知識所需的記憶歷程，以及在接受到一張影像所給予的訊息後，每個人所對於空間知識的記憶方式的異同。
本文使用模糊理論的成員函式表示空間中影像物件的關係，如此可以將影像上的訊息有效率的萃取成為空間的知識。為此，設計了一種空間知識的模型，用來表示空間知識中的方向與距離的不明確關係，建構符合人類由影像中編碼取得的空間知識。
在本文所設計的學習系統中，利用相似度與差異性的函式，比較在標準圖形與使用者所繪製的空間知識，在診斷的過程中，找出物體間錯誤的方向與距離的關係，並在診斷之後給予學習的建議，應證了之前所設計的空間知識模型，以期此系統達到學習的功效。

關鍵字

資料擷取 ； 空間關係 ； 空間知識 ； 模糊理論 ； 相似度 ； 影像物件

並列摘要

Abstract
In our daily life, people often use many kinds of spatial knowledge, among which spatial relationships can be interpreted through imagery. By images, learners can understand the places unreachable or never seen. Therefore, if a learner can make good use of this media, he or she can get various spatial relationships and knowledge of different areas without going out.
Starting from image information retrieval technology, this thesis extracts spatial information contained in images, such as the information of color attributes, texture patterns and shapes. The basic elements of spatial knowledge, called image objects, are then abstracted from the spatial information in an image. Thereupon, this study takes the advantage of image objects to acquire spatial knowledge, and looks into the characteristics of memory history in human brain when people recognize the messages transmitted by the image.
In our approach, the membership functions of fuzzy theory and image processing technology are adopted to investigate the spatial knowledge storage strategies of human brains. Such strategies can be simulated with elaborated data structure through computer. Furthermore, with the techniques of information retrieval, the information within the image can be extracted effectively. So the information and knowledge of spatial relationship can be reasonably transformed into concepts as well. As a result, a model of spatial knowledge is built up. With this model, those relations about uncertain directions and distances can be formulated to construct spatial knowledge.
In order to prove our model, a diagnosis system is designed with similarly function and the corresponding learning-diagnosis method is provided accordingly. The diagnosis system can detect spatial knowledge of learners and can then propose some learning advices. Experiment shows that such learning environment can help students learn spatial knowledge effectively.

並列關鍵字

similarity ； spatial knowledge ； information retrieval ； spatial relationship ； fuzzy theory ； image object

參考文獻

[App70] Appleyard,D.(1970).Style and methods of structuring a city. Environmentand Behavior,2,100-118

[Baj73] Bajscy, R. (1973). Computer identification of visual surfaces. Computer Graphics and Image Processing,2:118-130

[BaW92] Basri, R. & Weinshall, D. (1992). Distance Metric between 3D Models and 2D Images for Recognition and Classification.AI Memo 1373,AI Laboratory, MIT, July

[BeK91] Berlin, B., & Kay, P. (1991). Color terms: Their universality and evolution. Berkeley, CA: University of California Press

representation of spatial relations: A familiar environment.Journal of