数据规范化算法在基础矩阵估计中的应用

Application of Data Normalization Algorithm to Estimating Fundamental Matrix

在研究整体最小二乘法的基础上,分析摄像机成像误差的统计特征,明确误差对基础矩阵估计算法的影响.针对所建立的误差模型,分析数据规范化算法在基础矩阵估计中的应用.在不同的噪声强度下,通过实验说明非等向性数据规范化算法能够有效降低误差对求解精度的影响,使基础矩阵线性估计算法具有较高的精度和稳定性.

This paper analyzes the statistical characteristics of camera imaging error based on studying total least squares, and clarifies the error effect on estimating fundamental matrix. On the basis of the error model, the paper also analyzes the application of data normalization algorithm to estimating fundamental matrix. Experiments show that anisotropic data normalization algorithm can reduce effectively the influence of error on the precision of solution with various noise intensities, thus making the fundamental matrix linear estimator algorithm more accurate and stable.