【摘要】 通过融入方向信息修改了仅基于梯度幅值的蛇模型定义 ,从而产生了新的动态轮廓线模型 .新模型可以明显地削弱虚假强边缘的影响 ;基于动态规划方法 ,给出了基于小波分解的多分辨率能量极小化算法 .由于结合了Mallat的小波边缘检测理论 ,该算法可以明显地降低时间消耗 ,在选择滤波器方面更具有柔性 ,而且降低了初始位置和参数敏感性 ,从而扩大了目标捕获区域并使得参数的调整得以简化 .更多还原

【Abstract】 The magnitude-only external force is modified by integrating direction information explicitly and a new snake model is derived. The new snake model can greatly eliminate the disturbance of false strong edges. Based on dynamic programming approach to snake model, a wavelet-based multiscale algorithm is proposed. Combined with Mallat’s wavelet-based edge detection theory, the proposed multiresolution algorithm can greatly decrease the amount of computation, become more flexible in selecting filter than continuation method, and has less initial position sensitivity and parameter sensitivity, which enlarges the target capturing region and makes the setting of parameters simplified.更多还原