透過您的圖書館登入
IP:13.58.82.79

摘要


細化是由影像中萃取線形之常用步驟,本研究探討中軸轉換法、剝皮法、及吹氣球法等三類細化演算法之優缺點。其中剝皮法包含Stentiford & Mortimer [1983]、Zhang & Fu [1984]、及Holt, et al. [1987]等方法。從是否能夠克服典型細化問題、細化後是否形變或造成斷線、及有無良好的執行效率的角度出發,探討各種方法的優缺點,並提出可能的解決方案。本研究以航照影像經邊緣線偵測萃取之影像及網格向量化之掃描影像進行測試。就計算效率而言,Zhang & Fu [1984]所提方法執行效率最高,最差者為中軸轉換法,就細化品質而言,吹氣球法細化後易產生偏移,中軸轉換法斷線情況嚴重;而剝皮法中,以Zhang & Fu [1984]方法為基礎,配合Holt處理鋸齒狀像元之方法(稱為Z-S+Holt法)對細化常遭遇的問題可獲得較佳結果;此外,由細化差異分析結果顯示,Stentiford法雖與Zhang & Fu [1984]為基礎之方法無明顯差異,惟Stentiford法在一些交叉點位置會產生一些小迴圈;從保留文字辨識中重要筆畫而言,以Z-S+Holt法之處理結果較佳。

關鍵字

骨架化 線圖徵擷取

並列摘要


This research discusses the advantages and disadvantages of three kinds thinning algorithms. These algorithms include the Medial Axis Transform, Ballooning Approach and Peeling Approach. In the Peeling Approach, there are several different schemes proposed by Stentiford & Mortimer [1983], Zhang & Fu [1984] (the Zhang & Suen method), and Holt, et al. [1987]. This study investigates different schemes from the aspects of overcoming classic thinning artifacts, geometric distortion, break-line problems, computational performance, and finally proposes probable solutions to these problems. Numerical experiments include scanned maps for vectorization, and scanned aerial photographs. In general, Peeling scheme proposed by Zhang and Fu [1984] provides the best efficiency MAT gives the worst. For the thinning quality, Ballooning method suffers geometric distortion, and MAT has serious break-line problem. Experiments indicate that Peeling based method, e.g. Z-S and Holt combined schemes Can handle thinning problems very well. But Stentiford scheme has several thinning artifacts. The best result can be derived with Z-S and Holt combined scheme.

被引用紀錄


張孝慈(2015)。上消化道診斷之超解析度放大技術之開發〔碩士論文,中原大學〕。華藝線上圖書館。https://doi.org/10.6840/CYCU.2015.00142

延伸閱讀


國際替代計量