中文摘要：

由于传统的Chan—Vese模型无法分割多个同质区域的目标，多相水平集方法会产生区域重叠问题。提出了多相图像分割的多尺度变分水平集方法，使图像各相互相独立，避免分割区域的重叠和漏分。同时，利用小波变换将图像分解成多尺度的逼近子图像，在子图像上进行图像分割，采用插值法将粗尺度上演化曲线投影到细尺度上作为初始轮廓线，透层分割直到原始尺度图像，有效抑制噪声并提高了计算速度。实验结果表明，该方法比传统的方法能更快速有效地分割图像。

英文摘要：

The classical Chan-Vese model can＇t segment homogeneous objects with multiple regions. The muhi- phase level set method for image segmentation can make the segmented regions overlap. In order to overcome this limitation of these methods, multi-scale and muhiphase level set method is proposed. The use of curves required for the segmentation of N regions and each curve represents one region. In the meanwhile, wavelet transform is used to get multi-scale images and segmentation is performed firstly in finer scale sub-image for each scale. Then the resulting curve is interpolated from coarse to finer scale until the original scale image is reached. The method can increase the robustness of the method to noise and reduce the computational cost. Experimental results of im- age verify that our model is efficient and accurate.