【摘要】 提出了一种在平移情况下通过计算图像的几何静矩求形心方式估计基础矩阵的算法。首先对通过平移方式所采集的两幅图像进行相关的图像处理,并用形态学图像处理方法滤除残余的孤立噪声点,然后提取出标定所需的角点,并运用力学中的几何静矩方式求出由不同数量的角点所组成的各种多边形的形心,然后用归一化的8点算法求出基础矩阵。经过大量的实验证明该算法实现了基础矩阵的高精度的线性估计。更多还原

【Abstract】 A fundamental matrix estimation algorithm has been put forward through computing the image geometrical static moment,which is determined by calculating the polygon centroid through image translation.Three setps are included.Firstly,two images acquired through translation are processed preliminarily and a morphological-based filter is used to eliminate the residual noise points.Secondly,the corners needed for calibration are detected,and various centroids of all kinds of polygons composed of different number of corners are determined.Finally,the fundamental matrix could be calculated with normalized eight-point algorithm.Plenty of experiments show that high-precision linearity estimation could be realized using this method.更多还原