基于小波变换的自适应梯度边缘检测算法

针对传统的单一边缘检测算法抗噪能力差、边缘不连续等不足,本文提出采用两种算法相结合的方式来进行边缘检测。首先,对原始图像进行多层小波分解;然后,对小波分解后的图像低频部分用提出的8点邻域自适应梯度算法进行边缘检测,依靠边缘生长方法保证检测出的边缘的连续性,对高频部分用小波变换的局部模极大值算法检测图像的边缘;最后,将各层边缘信息按一定的融合规则融合起来得到最终的图像边缘。实验结果表明,该方法与传统的边缘检测算法相比具有定位精度高、去噪效果好等明显的优点,也能较准确地提取图像的边缘。

An Adaptive Gradient Edge Detection Algorithm Based on Wavelet Transformation

The image edge detection is one of the basic contents in image manipulation and analysis.Against the deficiency of traditonal single edge detection algorithms,e.g.,low anti-noise capacity,discontinuous edge and etc.,this paper proposes a new edge detection method combing two algorithms.Firstly,the original image is transformed by multi-layer wavelet decomposition to obtain respective approximate low frequency and detailed high coefficients.Secondly,for the low frequency part of wavelet decomposition image,an eight-point neighborhood adpative gradient algorithm is used for edge detection,and edge growing is used to ensure edge continuities.For the high frequency part,a wavelet transform partial max algorithm is used to detect the image edge.Next,the edge information of all the layers are combined as a certain rule to get the final edge of image.The results indicate that compared with the traditional edge detection algorithms this proposed method has the advantages of higher precision and signal-to-noise ratio,and it can make image edge extraction more accurate.