基于SIMD-MCC的2D图像矩并行算法

在模式识别和计算机视觉中，不变矩发挥着很重要的作用，因为图像在经过平称、旋转和伸缩变换后具有力矩不变的良好特性，直接计算二维图像力矩，需要大量的加法和乘法运算，而乘法运算最为费时。文中提出了一个在ＳＩＭＤ－ＭＣＣ计算机上实现的计算二维图像矩的并行算法。它的基本思想是将一个二维图像矩分解成很多个垂真矩和一个水平矩，利用数据并行性在ＳＩＭＤ－ＭＣＣ计算机上进行计算。这个算法极大地利用了计算的并行性，其

A parallel algorithm for the computation of two-dimensional image moments on SIMD-MCC

Moment invariants play an important role in pattern recognition and computer vision. They have the nice properties of being invariants under translation, rotation and scaling. To compute moments of a two-dimentional image, a large number of multiplications and additions are required in a direct approach. Multiplications, which are the most time-consuming operations in simple processors, can be completely avoided in the proposed algorithm for low order moments. In this paper, we develop a par allel algorithm for the computation of two-dimensional image moments on SIMD me sh-connected computers. The basic idea is to decompose a two-dimensional momen t into many vertical moments and a horizontal moment, and to use the data parall elism for the vertical moments and horizontal moment. The algorithm exploits par allelism of the computation maximally and has an idea time complexity of O( log 2 N) . Compared with algorithms available, it has a great improvemen t on computing speed.