頁籤選單縮合
題 名 | 應用橢圓參數於工業視覺之辨識 |
---|---|
作 者 | 戴國圓; | 書刊名 | 技術學刊 |
卷 期 | 9:2 1994.06[民83.06] |
頁 次 | 頁245-251 |
分類號 | 440.8 |
關鍵詞 | 工業視覺辨識; 電腦視覺; 橢圓參數; 傅立葉描述元; 影像處理; Industrial visual inspection; Computer vision; Ellipse parameters; Fourier descriptor; Image processing; |
語 文 | 中文(Chinese) |
中文摘要 | 本論文應用電腦視覺技術提出一個工業視覺辨識系統,它主要包含四個步驟:(一)受測物件圖像模型建立,本模型以傅立葉級數作為物件的描述。(二)受測物件影像前處理,包括雜訊的濾波,影像二元化處理,及受測物件邊界萃取。(三)受測物件傳立葉描述元的產生,(四)與圖像模型中的資訊作辨識。本系統以個人電腦為基礎,並配合影像處理卡(VFG-512)及CCD攝影機。已有四組工業圖像分別以各種位置,大小及方向來測試,結果發現本系統辨識正確率高達97%。而且整套系統能應用在個人電腦上執行,使得本系統大大提高它在實際應用上的價值。 |
英文摘要 | In this paper, a computer visìon system is proposed for industrial visual inspection. The system consists of four steps. First, the workpiece model is built by modeling the boundary of workpieces by Fourier series. Secondly, the image pre-processing which includes median fìlter, image binarization, and boundary detection is proceeded. Thirdly, a fourier descriptor of the input pattern is generated. Finally, the matching between the ìnput workpiece and models in database is evaluated using the least-square error method. The system has been implemented on a personal computer. The image data are captured by a visual interface board (VFG-512) and a CCD camera. In the experimental results, the success rate of recognition in this system is about 97% for four kinds of industrial patterns with various translation, scaling, and rotation situations. As a result, we have proposed a computer vision system for industrial visual inspection successfully. |
本系統中英文摘要資訊取自各篇刊載內容。