摘 要:現存的基于空間域的圖像邊緣檢測算法只能有效檢測出圖像有限方向的邊緣。針對這一問題,根據Contourlet變換子帶的各方向子帶代表的方向信息及其梯度方向,提出了一種基于Contourlet變換和Susan算子的邊緣檢測算法。邊緣檢測算法首先對源圖像進行Contourlet變換,然后分別對高、低頻子圖像進行邊緣提取,最后通過一定的融合規則進行融合,得到邊緣圖。實驗結果表明,這種邊緣檢測方法具有有效地抑制噪聲、邊緣檢測精度高等特點,是一種有效的圖像邊緣提取算法。關鍵詞:邊緣檢測; Contourlet變換; Susan算子; 空間域
中圖分類號:TN911-34;TP391 文獻標識碼:A
文章編號:1004-373X(2010)16-0137-03
Image Edge Detection Based on Contourlet Transform and Susan Operator
LI Jie, XIANG Jing-bo
(China Airborne Missile Academy, Luoyang 471009, China)
Abstract: The available image edge detection algorithm based on the spatial-domain can capture only limited directional edge in image. An improved algorithm for the image edge detection based on the directions of each directional sub-band and its gradient is proposed to overcome the disadvantages. By the algorithm, firstly, multi-scale decomposition of the image is performed with Contourlet transform. Then the edge detection of low-frequency sub-image and high-frequency sub-image are obtained. Finally, this edge diction method combines the advantages of both methods together to fuse the two edge information obtained by different methods according to certain rule. Experimental results show that the new edge detection method is effective to suppression of noise, and has advantages of high precision image edge detection.Keywords: edge detection; Contourlet transform; Susan operator; spatial domain
收稿日期:2010-03-30
基金項目:航空科學基金資助項目(20090112004)
邊緣是圖像最基本的特征,圖像的邊緣定義為周圍像素灰度變化不連續的像素點的集合,也就是圖像中具有奇異性的像素點的集合。邊緣檢測是圖像分析、目標識別以及圖像濾波的前提和基礎。
圖像信號的奇異性通常包含了圖像最本質的信息,圖像中的邊緣是圖像中的奇異點的集合,邊緣檢測算法則需要發現這些奇異點并對其進行準確定位,加強圖像的輪廓特征,以便于人眼和機器的識別。
傳統的圖像邊緣檢測方法有Prewitt 算子、Sobel算子、Roberts算子、Kirsch算子、 Canny算子、Log算子等[1-4]。但這些算法以一階導數極大值點或二階導數過零點作為候選邊緣點,通過選取合適的閾值,從中提取圖像邊緣,因為微分運算對噪聲比較敏感,不能抑制噪聲,常把噪聲當作邊緣點檢測出來,而真正的邊緣卻可能被漏檢。……