機器視覺系統主要是在影像上進行特徵點偵測,並透過這些特徵點進行分析與應用,如尺寸檢測、瑕疵檢測、精密定位等。而本研究以機器視覺為基礎開發一套特徵重建演算法,透過CCD元件擷取物件影像,經過小波去噪以及解析度調整,搭配SLIC超像素分割法將影像內的各封閉曲線分離出來。接著使用形態學處理取得物體的外型輪廓,並提出一種以向量循邊為基礎的特徵點偵測法來進行特徵點辨識,利用特徵點分割物體外型輪廓,將輪廓進行線段分類與重建,使我們獲得物體的尺寸資訊及輪廓特徵資訊,希望能讓使用者藉由這些數據,進行工具機加工數據回饋及統計分析。在後續也針對不同情況下的特徵點辨識能力作探討,與一般常用角點偵測算子做比較,在不同影像解析度與旋轉角度下,本研究所開發之特徵點偵測法依然可正確的完成偵測。而我們也取三種不同複雜度外型的物件來進行辨識與重建,皆能達到正確的重建結果。
The machine vision system focuses on detecting feature points on the image as well as analyzing these feature points for various applications such as size detection, defect detection, precision positioning, and so on. The main purpose of this paper is to develop a feature reconstruction algorithm based on machine vision. First of all, we capture the object image by CCD component, undergoing the process of wavelet denoising and resolution adjustment. Via SLIC super-pixel segmentation, we separate the closed curves in the image, and then obtain the contour of the separated objects by means of morphological processing. In this study, we propose a feature point detection method based on vector edge tracing. The contours of our concerned objects are segmented via feature points. Then these contours are categorized and reconstructed so as to obtain the size and contours information. We hope users can utilize these data in machining data feedback and statistical analysis. Next, we discuss the feature point detection ability for our method under different circumstances. In comparison to commonly used corner detection operator, our method can still correctly complete the identification under different image resolutions and rotation angles. Besides, we take three objects with different complexities to identify and reconstruct via our methods. All the results are correctly reconstructed.