摘要
针对主动轮廓模型(Snake)处理图像时的初始轮廓选取问题,采用改进的区域增长法对图像进行预分割,并将得到的边缘作为主动轮廓模型的初始边缘轮廓;然后分别用sobel算子与梯度矢量流(GVF)代替图像梯度进行主动轮廓模型外部能量的计算,在速度满足要求的情况下,提高了目标区域的提取精度。实验结果表明,基于GVF的主动轮廓模型在脑肿瘤的轮廓提取中能取得更好的效果。
For selection of original object contour to active contour model, the improved region grow algorithm is adopted for pre-segmentation in this paper, then the extracted object contour is taken as initial contour of active contour model for precise segmentation computation. And in the calculation of Eext of active contour model, it uses GVF instead of gradient, therefore, the operation time is under control, while the accuracy of ROI detection is improved. The experimental results show that the improved active contour model based on GVF can get a better result for contour detection of MRI brain tumor.
出处
《计算机与数字工程》
2007年第11期89-92,共4页
Computer & Digital Engineering
