基于小波变换的多尺度图象边缘检测

The Application of Wavelet Multi-Scale Transformation in Image Edge Detection

边沿作为图像视觉的最主要特征 ,成为图像信息获取的重要内容 .而小波变换具有检测局域突变的能力 ,而且可以结合多尺度信息进行检测 ,因此成为图像信息边缘检测的优良工具 .基于信号与噪声在不同尺度下小波系数模的变化特征 ,利用小波变换系数模局部极大值来提取图像的边缘特征 .实验结果说明这种特征提取方法不仅有效地降低了噪声 ,而且也能较准确地提取图像的边缘及降低计算量 .

The border becomes the important content that information of the picture is obtained as the main characteristic of the vision of the picture. Wavelet can describe the change of image better than others, and it can combine the multiiscale method to measuring. So it become the fine tool for the pictue processing. Based on fifferent properties of the signal and noise in the wavelet coefficient modulus under the different scale of wavelet transformation, an edge detection method is proposed. Detecting the local maximum of the wavelet transformation coefficient modulus, it provides information of the image boundaries feature. Using the property we can detect image edge.Experimental result shows that the method on\t only eddectively reduces noise, but also accurately extracts the image edge and has less computational time.