基于结式法计算目标的空间方位

Computing the Problem of Object's Orientation by Resultant Methods

根据图像信息确定立体目标的空间方位是计算机视觉中的一个基本问题。利用图像亮度的二维矩将该问题可以化为七个六变元的非线性多项式方程组。此文提出利用结式法求解该方程组的闭形式解。该方法一次可以消去多个变元,克服了传统方法中逐个消去变元的弊端,提高了算法的效率。除此之外,我们的方法是确定的构造性算法,并且发挥了结式理论在符号消元方面的优势。因而,该方法高效地解决了目标空间定位中的相关计算。

Relative orientation is a basic problem in computer vision.This problem can be transformed to 7 polynomial equations with 6 variables by using two dimensional moment function of the image lightness.Resuhant method is presented to solve the above equations.Unlike the classical algorithm,in which one has to eliminate the variable one by one,our method can eliminate three （or more）variables simultaneously.Moreover,it is constructive algorithm and takes the advantage of symbolic elimination of resultants.Therefore,it is very efficient in solving the problem of relative orientation.