描述了有噪图像边缘检测的方法，它基于小波变换的多尺度分析。通过合理选择小波基，可以得到图像的多分辨表示。证实了小波多尺度变换在有效去噪的同时可以保持图像的特征。通过信号，噪声奇异性的分析，对有噪图像小波变换局部极大值的检测可提取出所需的图像边缘特征。实验结果证实了该方法的可行性 The method of edge detection of noisy images is described, which is based on the multi-scale analysis of wavelet transform. By choosing the wavelet basis reasonably, the multi-resolution representation of the image can be obtained. It is proved that wavelet multiscale transform can preserve the features of the image while effectively denoising. Through the analysis of signal and noise singularity, the required image edge features can be extracted by detecting the local maxima of the noisy image wavelet transform. Experimental results confirm the feasibility of this method