基于标记的Watershed图像分割算法是一种改进的分水岭算法，它很好地被应用于人脸及其他一些场景图像的分割。把该方法引入高分辨遥感图像的分割，并针对遥感图像的特点，在分割之前采用中值滤波对高分辨率遥感图像进行预处理；同时，在分割过程中采用用小波滤波器替代Butterworth滤波器对梯度图像的低通滤波。不同地物特征Quickbird图像的分割实验表明，对于纹理比较均一的高分辨率遥感图像，该方法避免了过分割现象，且效率较高；但对于纹理比较复杂的图像，该方法具有一定的局限性。-Watershed based image segmentation algorithm tag is an improved watershed algorithm, it well be applied to the face and some other scenes segmentation of images. The introduction of high-resolution remote sensing of the method for segmentation of images and features for remote sensing image, the segmentation using median filtering before the high-resolution remote sensing image is preprocessed　also used in the segmentation process Butterworth filter with a wavelet filter replacement the gradient image of low-pass filter. Characteristics of different objects that Quickbird image segmentation, for texture and uniform high-resolution remote sensing images, the method avoids over-segmentation phenomenon, and efficient　but more complex texture images, this method has some limitations .