兼具時空向度及物體型態特徵之行動檢索研究

A Mobile Retrieval Study on Joint Estimation of Temporospatial Geo-information and Bird Morphological Features

王凱弘(Kai-Hung Wang)

指導教授 ： 王獻章

長榮大學/資訊暨工程學院/資訊管理學系(所)/碩士(2011年)

https://doi.org/10.6833/CJCU.2011.00196

摘要

在進行物種的查詢時，一般主要以外型特徵為關鍵，而其時間與空間屬性亦可能具有顯著性的影響。進行物種辨識時，若考量時空向度將可提升查詢之正確性。本研究目的為提出一套兼具時空向度與物體型態特徵之鳥類知識檢索系統，並建置於行動裝置以提升使用的便捷性。
研究方法主要包含：(1)分析及建立賞鳥記錄與鳥類知識庫：提供後續查詢與呈現之使用；(2)建立特徵部位選擇介面與最大似然機率模型於受觀測鳥種；(3)開發符合本研究需求之使用者操作介面與實驗環境：進行實驗結果收集與評估。
實驗部分，將查詢方式分成三種(1)不考量時空向度；(2)不同時間；(3)不同地區，並探討其Top-N正確率的差別。結果顯示，若未考量時空向度條件，容易因比對相似結果過多，照成無法在介面上呈現目標鳥種；而考量時空向度後，將可提升Top-N的正確率。此外，本研究經卡方檢定歸納出5個較具代表性的部位，可用於介面選單設計，加快使用者對介面的選擇與操作效能。

關鍵字

時空向度、特徵查詢、最大似然機率、卡方檢定、鳥況記錄、Top-N正確率

並列摘要

Objects queries generally based on shape features, and for some specific objects, time and space properties have significant influence. While recognizing object by shape characteristic, the consideration of space-time dimension will increase the correct rate of queries. This research aims to propose a set of both time and space dimensions and characteristics of the object type of bird knowledge retrieval system and build on the mobile device to enhance ease of use.
The major research methods includes:1) Analyze and establish the bird-watching record and knowledge databases :To provide follow-up queries and presentation. 2) Build the characteristics selection interface and the maximum likelihood probability model based on specific birds. 3) Development meet the needs of the user interface research lab environment: For collection and assessment of experimental results.
The experiments are divided into three parts according to the parameters used in the query: (1)Does not consider the temporospatial parameters. (2)Consider different temporal parameters. (3)Consider different spatial parameters. The results showed that, if not consider the spatial and temporal dimensions of conditions susceptible to excessive compared to the similar results, according to the interface can not be present the target bird species; and consider the spatial and temporal dimensions, the Top-N will improve the correct rate. In addition, this study summarized by the chi-square test more representative of five parts, the menu can be used to interface design to speed up the user interface of into choice for performance and operation.

並列關鍵字

temporospatial parameters ； feature query ； maximum likelihood estimation ； Chi-square test ； bird record ； top-N

參考文獻

[5] 李政傑(2008)，「物體描述之共通性分析與應用-以鳥類之口語化描述語料為例」，長榮大學資訊管理研究所論文。

[16] 楊佩菁(2006)，「專業語料與模糊描述中物體特徵自動擷取及比對之研究」，長榮大學資訊管理研究所論文。

[21] 蔡文吉，李嘉紘(2007)，「利用影像特徵擷取之貝類檢索系統」 ，崑山科技大學碩士論文。

[23] Chen, Y., Kao, T., and Sheu, J.( 2003) ,“A mobile learning system for scaffolding bird watching learning.” J. Computer-Assisted Learning 19 (2003), 347--359.

[24] Chih-Hsun Chou, Chang-Hsing Lee, Hui-Wen Ni,(2007) “Bird Species Recognition by Comparing the HMMs of the Syllables,” icicic, pp.143, Second International Conference on Innovative Computing, Informatio and Control (ICICIC 2007).