Design and Implementation of an Intelligent Game Engine Framework for Natural Interaction Enabled Multiplayer Games
| Autor: | Chiu Chih-Ming, 邱志明 |
|---|---|
| Rok vydání: | 2014 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 102 In this paper, an intelligent game engine framework called GENI is proposed to recognize, analyze, and predict player actions on-line. First, the martial art styles are recorded using the Kinect device, then we utilize the Dynamic Time Wrapping (DTW) algorithm to analysis and recognize real-time user input style. Second, after learning the model, we collect the logs of the on-line recognized behaviors data, extract interesting patterns, and obtain reliable probabilities from the mining step. Two efficient approaches namely MSSBE and MSSMB are proposed for mining the behavior data and finding the interesting style patterns. In addition, a suffix matching algorithm is proposed to discover the proper style sequences of an avatar for predicting future responses of opponents. The N-Gram predictor algorithm is also presented for evaluating the efficiency of our methods. Finally, to demonstrate how those algorithms apply to real game scenarios, we implement the framework using a scene graph based approach with rich graphical user interfaces to fulfill the game experiment environment. Moreover, we also apply GENI in walkthrough research and propose several new approaches based on a behavioral oriented system that uses traversal patterns to model relationships between users and exploits semantic-based clustering techniques, such as association, intra-relationships, and inter-relationships, to explore additional links throughout the walkthrough system. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|