- 學位論文
一個有效率的二階形態學運算方法
An efficient method for binary morphology operations
摘要
數位影像處理研究的重要目標之一是為圖像處理領域發展更有效率的運算方法。近年來，形態學的圖像處理技術受到了許多關注，因為它包含圖像數據的簡化、圖像對基本形狀特徵的保存及消除不必要的內容，更重要的是，可以基於基本的形態學運算來建構複雜且有效的圖像處理任務。儘管其理論非常簡單、易於理解及實現，但往往存在運算速度慢的缺點，不經改善難以實際使用。 在本論文中，我們提出了一種Run-Based的圖像表示法應用在二階形態學的運算當中，我們定義一個run為：在每一行(colomn)中連續像素值為1的像素序列，用來表示目標圖像以及結構元素，並實現了基於此表示法進行膨脹、侵蝕運算的技術和算法。 原始方法的膨脹、侵蝕運算時間複雜度皆為O(N^2 M^2)；而本文所提出的Run-Based表示法時間複雜度為O(NM(log(M)))和O(NM)。實驗結果表明，使用本文所提出的方法可以加快運算的時間。
並列摘要
One of the important goals of digital image processing research is to develop robust and efficient algorithms for many image processing tasks. Over recent years, mathematical morphological image processing has received a lot of attention since it includes the simplification of the image data, the preservation of essential shape characteristics for image objects and the elimination of irrelevancies. More importantly, complex and useful image tasks can be constructed based on the basic morphological image operations. Although its theory is very simple, easy to understand and implement, it often has the disadvantage of slow operation speed. It is difficult to use it without improvement. We propose a Run-Based image representation applied to binary morphology operations. Run-Based image representation is an alternative way of representing a binary image and the structure element using a run, which is a sequence of ‘1’ pixels on the columns. In this paper, techniques and algorithms for performing morphological image processing based on the proposed approach are presented. The time complexity of the dilation and erosion calculations of the original algorithm is O(N^2 M^2); and the time complexity of the dilation and erosion calculations proposed in this paper is O(NM(log(M)) and O(NM). Experimental results demonstrate that a speed up in computation time can be obtained using the proposed method.