頁籤選單縮合
題名 | 以影像處理與類神經網路分級胡蘿蔔之研究=Carrots Grading with Image Processing and Neural Network |
---|---|
作者 | 黃膺任; 李芳繁; | 書刊名 | 農業工程學報 |
卷期 | 43:3 1997.09[民86.09] |
頁次 | 頁86-101 |
分類號 | 435.213 |
關鍵詞 | 影像處理; 類神經網路; 胡蘿蔔; 分級; Image processing; Neural network; Carrots; Grading; |
語文 | 中文(Chinese) |
中文摘要 | 本文主要目的是結合影像處理技術與類神經網路來建立一套胡蘿蔔的分級系統。 其分級方式, 首先以間隙追蹤法找出影像邊界,以 Hotelling 法找出影像長軸,再根據長 軸及影像邊界的相對關係萃取出對稱率、直徑差異度、曲率及細密度等描述形狀的特徵因子 ,並以根肩綠色與紅色在色度圖上有不同分佈的特性,找出屬於綠色像素的判斷模式,藉以 計算根肩綠色面積來作為描述顏色的特徵因子,結合形狀及顏色因子作為分級參數,並配合 一個三層式倒傳遞神經網路來模擬人工分級作業,同時藉由改變輸入參數種類,數目及隱藏 層節點數,以尋求分級胡蘿蔔之最佳模式。實驗結果,由神經網路所建立的分級模式與人工 分級結果比較,其相符程度為 81%。 |
英文摘要 | Image processing techniques and artificial neural network were used in this study to establish a grading system for carrots. At first, the boundary pixels of the carrot image were located using the crack-following method, and the major axis was obtained using the Hotelling transform. Then, the shape descriptors-the symmetry ratio, the diameter difference, the curvature, and the compactness were extracted based on the relative position between the major axis and image boundary. The green area of the root-shoulder portion, used as the color descriptor, was calculated by utilizing the chromaticity diagram. A three-layer backpropagation artificial neural network was utilized to simulate human grading operations. The shape and color descriptors were used as the input parameters of this neural network. In order to find the best model for grading, the type and number of the input parameters, the number of the nodes of the hidden layer were changed during the training process. The grading accuracy of the best model was 81% compared with human grading results. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。