الفهرس 
Fundamental ConceptsOnly 14 pages are availabe for public view
 |  | from | 175 |  |  |
| | | from | 175 | | |
Abstract
Sketching facilitates the process of investigating and clarifying dia- grams, schemas, plans, charts, and other coarse organizations. Freehand sketching interfaces can provide an electronic alternative to the classical pencil-paper means to connect ideas and designs to easily create, record, or modify them as needed. But the available interfaces do not provide, yet, a representation formalization of the aspects of hand-drawn sketch- ing. This motivates us to design a new technique for Recognizing Sketches (RS) components (strokes). Thus, the objective of the RS technique is to teach electronic machines to understand what is being sketched on the interfaces. The RS technique can be applied to any set of shapes and any set of real-valued features to recognize a single-stroke drawing. In this thesis, it is applied to eight geometric shapes that could be lines, circles, ellipses, triangles, quadrilaterals, polygons, polylines, or curves.
The RS technique utilizes global geometric features, Fuzzy Formal Concept Analysis (FFCA), and a similarity function. It provides an im- plemented interface that allows users to draw strokes. Then, the recog- nition result is retrieved as one of the aforementioned geometric shapes. The implemented system uses real-time stroke recognition, so it could be utilized in many artificial intelligence applications like “face recognition” and “sketch design software”.
This thesis contains six chapters, an appendix, and a list of references. Its structure goes as follows. Chapter 1 gives an introduction about the thesis, where the motivation, the objective, the contribution, and the or- ganization of the thesis are presented. Chapter 2 introduces fundamental definitions and basic concepts of partial order and lattice theory, strokes and geometry, Formal Concept Analysis (FCA), and FFCA. Chapter 3 presents some related work including previous sketch recognition systems and the related work of the main project1, of which the thesis originally emerged. In addition, some applications that utilize FCA or FFCA tools are given. Such tools are also presented in this thesis.
The proposed RS technique consists of five phases that are demon- strated in Chapter 4 and Chapter 5. Chapter 4 includes the experiment illustration, the outlines of the RS technique, and the first two phases. Phase 1 is devoted to extract features from the experimental data. In Phase 2, an algorithm for extracting a salient feature for each shape is explained.
Chapter 5 contains Phase 3, Phase 4, and Phase 5 and also gives recognition results. A suggested algorithm called SILFR-algorithm in- vestigating a sequence of features for each geometric shape is explained in Phase 3. Each sequence is able to recognize a corresponding stroke. While, Phase 4 presents a new methodology of applying FFCA to re-
1JESICS project which has been funded by the German Academic Exchange Service (DAAD) within the framework of the German-Arab Research partnerships Programme Line 4 under grant agreement 57247603.
formulate the distinguishing sequences of features which resulted from Phase 3. The usage of FFCA decreases the number of computed features for recognizing a new stroke. Finally in Phase 5, a new similarity function is utilized for recognizing the strokes that are not recognized by FFCA. In the recognition results section, the RS technique is evaluated, and the accuracy of the RS technique is shown. Chapter 6 gives a conclusion, the achieved results of the RS technique, and a suggested future work. At the end of the thesis, Appendix A provides important parts of the RS technique implementation.