中文摘要：

为了更好地利用snake模型来提取彩色图像中的物体轮廓，提出一种改进的snake算法。此方法首先自动生成snake的初始模型，然后在GVF-snake的基础上重新设计了snake的外部能量函数，采用色彩聚类算法对原始图像进行分割，利用像素到聚类中心的距离增强图像并进行差分运算，提取有意义区域的边缘梯度，对GVF向量场进行了归一化处理并改进了平滑因子。实验结果证明，改进后的算法，特别是在处理彩色图像时，大大优于原始方法，提高了轮廓提取的精度且有较好的鲁棒性。

英文摘要：

In order to improve the effectiveness of snake active model to extract the object contour in color irnages, introduce an improved snake algorithm. First, develop a new approach to generate an initial contour automatically; then a new external force is redesigned -basing on the GVF - snake. Use the color - based clustering algorithm to divide the image. Then use the distance between the pixels and the clustering center to enhance the image and use the difference to get the edge grads of interest. In addition, normalize the GVF field and improve the smooth factor. Comparing the snake model prototype with the new method,it indicates that the improvement is considerable, especially when color images are involved.