基于提升算法的不可分离小波图像边缘检测

Detection of the image edges of non-separatedwavelets based on lifting scheme

针对传统可分离小波图像边缘检测方法的各向异性和计算复杂性,应用基于“五株排列”的提升不可分离小波的构造方法,把基于矩形栅格的数字图像分成红块和黑块,然后根据图像边缘特点,利用二维提升格式构造了“最大-提升”小波,进而提出了一种基于提升算法的不可分离小波图像边缘检测新方法,通过与传统的B样条小波边缘检测方法进行试验比较,证明了新方法在图像边缘检测方面具有更好的效果.

Traditional image edge detection of separated wavelet shows its anisotropy and computational complexity. A new method for image edge detection was proposed by using lifting scheme of nonseparated wavelet based on the quincunx lattice, in which an image consisting of rectangular lattices was split into the red and black subsets. According to the characters of the image edge, the max-lifting wavelet was constructed using the 2D lifting scheme. The new method was compared with the classical edge detection way using B-spline wavelets. The results showed that the new method is superior to the traditional one in the experiment.