以gradient vector flow與distance transform為基礎的中軸演算法

Abstract

中軸是一種用來描述物體形狀的方法，由於具有精簡化的結構且保留原物體的拓樸等特性，所以在很多方面都可拿來應用，如分析影像、圖形辨識等。但是由於中軸沒有明確的定義，所以計算上仍然是不容易的。本文提出一個結合Gradient Vector Flow與Distance Transform作為尋找中軸的方法，並且在具有不同特徵的影像上做實驗，也顯示結果符合大部分用來尋找中軸的條件。

Centerline is one of many methods to describe the shape of objects. The method simplifies and maintains the main structure of objects such as topological characteristics. Therefore, the preference of searching for centerlines in image analysis and pattern recognition is mostly applicable. However, centerline method is not well-defined and consequently causes the difficulties of calculation. The thesis provides a way to search for centerline of objects by integrating gradient vector flow and distance transform. In addition, we apply the method on the images with various figures and the result indicates great match with centerline conditions.