基于一类中心B样条二进小波的多尺度边缘提取

MULTISCALE EDGE DETECTION BASED ON ONE KIND OF CENTER B SPLINE DYADIC WAVELET

提出一类渐近收敛于高斯函数的导数的中心Ｂ样条二进小波，用以实现多尺度Ｃａｎｎｙ最优化准则．在应用Ｃａｎｎｙ准则的实际计算中，必须对高斯函数的导数进行截尾．文中通过分析开窗截尾高斯函数的导数，对滤波器性能和计算的快速和有效性的影响，提出了利用具有紧支集的一类中心Ｂ样条二进小波作为滤波器，一方面可以近似满足Ｃａｎｎｙ提出的最优滤波器的要求；另一方面。

In this paper it is proposed that using a kind of center B spline dyadic wavelet that converges to the derivatiove of Gaussian can realize multiscales optimal criterions which is propsed by Canny. In realizing the Canny opreator, we must cut off the tail of the derivative of Gaussian. In this paper we analyses how the opening window on Canny operator effected the caculation speed and it is proposed to use a kind of canter B spline wavelet as filters, which can satisfy the optimal criterions approximately and detect edges with very high caculation speed