English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 51296/86402 (59%)
造訪人次 : 8167632      線上人數 : 131
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library & TKU Library IR team.
搜尋範圍 查詢小技巧:
  • 您可在西文檢索詞彙前後加上"雙引號",以獲取較精準的檢索結果
  • 若欲以作者姓名搜尋,建議至進階搜尋限定作者欄位,可獲得較完整資料
  • 進階搜尋
    請使用永久網址來引用或連結此文件: http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/94520


    題名: 基於PSO模糊分類器於多目標色彩模型的設計
    其他題名: Pso-based fuzzy classifier in multi-object color models design
    作者: 林怡女勻;Lin, Yi-Chun
    貢獻者: 淡江大學電機工程學系碩士班
    李世安;Li, Shih-An
    關鍵詞: 色彩分割;模糊系統;粒子群最佳化演算法;全方位視覺;color segmentation;fuzzy system;PSO;Omnidirectional Vision
    日期: 2013
    上傳時間: 2014-01-23 14:44:05 (UTC+8)
    摘要: 本論文係以符合FIRA (Federation of International Robot-soccer Association) RoboSot規則之中型足球機器人全方位視覺系統為發展平台,研究基於PSO模糊分類器之多目標色彩模型的設計,取代傳統以人工建立多目標色彩模型的方法,提升建立色彩模型之效率,並且改善一般以HSV色彩空間中色相、飽和度與亮度之上下界建立色彩模型容易包含過多不屬於目標色彩範圍的問題,減少機器人誤判目標物的可能。本論文提出解耦合模糊分類器的方法簡化PSO訓練參數之複雜度,一個模糊分類器建立一種目標色彩模型,單一模糊分類器之輸入為像素之三分量-色相、飽和度、亮度,輸出為像素對應模糊規則庫之最大歸屬度與其所屬規則類別,綜合模糊分類器獲得歸屬度最大之所屬類別即為輸出。PSO模糊分類器為監督式學習方法,一般樣本空間以事先在多種環境下取得之多目標樣本組成,而本論文以固定位置作為基本定位,定義感興趣之多目標物範圍後就地取樣,建立適合當時環境之色彩模型。實驗結果呈現本論文提出之建立多目標色彩模型方法較使用六個閥值建立之色彩模型貼近多目標物之色彩模型,且提升人工建立色彩模型之效率。
    The construction of multi-object color models based on PSO-based fuzzy classifier is proposed. This study is developed on omnidirectional vision system of middle-size robots with the competition of FIRA (Federation of International Robot-soccer Association) RoboSot. It replace the old method in order to promote the efficient of construct multi-object color models by manual and improve the problem which misjudge the object because of using six thresholding in HSV color space may include too many colors not belong objective color model. This paper proposed a method to reduce the complexity of training parameter in PSO algorithm with a decouple fuzzy classifier. One fuzzy classifier will construct an objective color model. The input for fuzzy classifier is the consistence in one pixel, hue, saturation, value. The maximum membership value with the color class is output. PSO fuzzy classifier is a supervised learning method. Generally, the patterns of multi-object were obtained in multi-environment beforehand. This paper use fixed location to define the region of interest then get the patterns of multi-object pixels right there. And construct the multi-object models suit of the environment. The experiments show the multi-object color models construct by the proposed method between by manual use six thresholding is much close to the real multi-object color models and much efficiently.
    顯示於類別:[電機工程學系暨研究所] 學位論文

    文件中的檔案:

    檔案 大小格式瀏覽次數
    index.html0KbHTML244檢視/開啟

    在機構典藏中所有的資料項目都受到原著作權保護.

    TAIR相關文章

    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library & TKU Library IR teams. Copyright ©   - 回饋