Abstract: Medical images with fuzzy and non-uniform characteristics make it difficult to accurately extract target contour, aiming at this problem, this paper adopts an improved genetic algorithm to make active contour model finish boundary extraction. It adopts prepotent operator to keep inheritance of quantitative characteristic and select proper crossover operator. In the anaphase stage of the improved genetic algorithm, it solves its changeover from global optimization to local optimization. Experimental results show that it is anti-fuzzy with good robustness in extracting contour of targets.

Key words: active contour model, contour extraction, improved genetic algorithm

摘要： 针对医学图像的模糊性和灰度不均导致目标轮廓难以准确提取的问题，提出使用改进的遗传算法控制主动轮廓模型完成边界提取的方法。采用保优算子保留遗传性状，选择适当的交叉算子，在进化后期可实现由整体寻优到局部寻优的转变。实验结果证明，该方法在提取目标轮廓时抗模糊能力强、鲁棒性好。

关键词: 主动轮廓模型, 轮廓提取, 改进的遗传算法

CLC Number: 

• TP391