摘要
矩是基于区域的形状描述子,相对于基于轮廓的描述子例如傅立叶、链码描述子等,对于不连通的图像形状描述和对噪声的鲁棒性等方面有着更良好的性能。正交矩又可分为连续正交矩和离散正交矩,Krawtchouk矩是离散正交矩中的一种,和连续正交矩不同,基于离散正交矩本身的离散特性,更适合于对数字图像的处理。但同其他离散矩一样,Krawtchouk矩并不具备天然的几何不变性(旋转、缩放和平移),这也从一定程度上限制了Krawtchouk矩的应用。为使Krawtchouk矩得到更广泛的应用,对Krawtchouk旋转不变矩的构造进行详细分析和实验,比较出更适合用于浮游植物的Krawtchouk旋转不变矩。
Moments is a shape descriptor based-on region, it has better performance for description of unconnected-region-image and robust against noise comparing to those methods based on contour such as fourier descriptor and chain-code descriptor. Orthogonal moments can glancing be divided into two categories, continuous orthogonal moments and discrete orthogonal moments. Krawtchouk moraents is one of discrete orthogonal moments. By comparison, discrete moments is more suitable for digital image due to its discreteness. As other discrete-moments, Krawtchouk moments is not basic invariant with geometric transformation （rotation, scaling and translation）, and Krawtchouk moments is not widely spread used in this case. To solve this problem, gives the suitable rotation invariant Krawtchouk moments, by comparing different methods and experiments.
出处
《现代计算机》
2009年第10期61-64,共4页
Modern Computer
基金
国家自然科学基金(No.40627001)
