基于模糊规则自动获取的模糊主动轮廓模型

为了引入更多的不确定性先验信息来指导轮廓提取,首先提出了一种新的模糊主动轮廓模型(FuzzySnake),然后针对模糊主动轮廓模型中模糊推理规则难以确定的问题,提出了一种改进的遗传算法———多种群共生遗传算法,并用于相关模糊规则的自动获取,最后给出了用新的模糊主动轮廓模型从人工合成图像、加噪测试图像和实际的超声图像中提取目标轮廓的实验结果.实验表明,在模糊主动轮廓模型中利用多种群共生遗传算法进行模糊规则自动获取,不仅能有效克服模糊规则需向领域专家学习才能获取的局限性,而且能有效地降低传统模型对噪声的敏感性,使模型的参数调整得以简化.

A novel fuzzy snake with automatic fuzzy logic rule bases generation using genetic algorithms

A novel fuzzy active contour model(fuzzy snake),which allows for the integration of more uncertain a priori knowledge into an active contour model and an intuitive specification of the properties of an object's boundary is presented.This is accomplished by setting up two fuzzy inference systems that describe each of the fuzzy snake's segments and help to specify the values of image energy and uncertainty energy separately.In(order) to realize fuzzy logic rules generation automatically,the multi-population coexist genetic algorithm is proposed which is employed to generate all fuzzy logic rules in fuzzy snake.As a matter of practicality,the contour extractions are performed in the synthetic images and real cardiac ultrasound images using the novel model.Due to the integration of more uncertain a priori knowledge,the experimental results show that our method is more efficient than the original snake and has less initial position and noise sensitivity.