摘要
针对图像的Legendre正交矩计算量大和矩值求解过程中存在离散近似误差等问题,提出一种新的高精度快速计算图像Legendre矩方法.文中首先提出一种最大块优先分块策略,然后在此基础上,根据图像像素灰度值的取值特征将图像进行分块表示,以每个图像块为单位计算图像的Legendre矩.实验结果表明,与现有的快速算法相比,文中方法在保证矩值高精确的前提下,有效地减少了算术运算的次数,降低了计算复杂度,具有较快的计算速度.
In order to decreasing the computational time and avoiding numerical approximation which is involved in Legendre orthogonal moments,a new algorithm for high-precision and fast computation of Legendre moments on gray scale images is presented.A maximum image block first representation strategy（MBF） is proposed which is used to extract blocks of all intensity values.After extracting the blocks,the image can be redefined and computed using integral formula in terms of blocks of different intensities.The experimental results show that,as compared with the existing fast Legendre moment methods,the proposed method reduces the computational complexity of Legendre moments and has the high precision.
出处
《小型微型计算机系统》
CSCD
北大核心
2011年第6期1142-1146,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60573019)资助
广东省科技计划项目(2007A020100006-10)资助
关键词
LEGENDRE矩
最大块优先分块策略
快速计算
图像处理
legendre moments
maximum image block first representation strategy
fast computation
image processing
