投影距离的概率峰值效应及其在三维目标识别中的应用

在计算机视觉中，由于物体的几何特性随视线的变化而变化，从而给三维目标识别带来了困难．在设计识别算法时，关键是理解目标重要特征的变化，如角度、距离等的变化．在视线方向各向同性条件下，投影角度和距离的概率密度在其原值附近具有峰值效应，即投影角度及距离接近于原角度及距离．利用投影角度的联合概率函数可识别多面体．在已知距离时，为了便于识别目标，导出了投影距离的密度函数．实验结果证明了该方法的有效性．

Probabilistic Peaking Effect of Projected Distance and Its Application to 3 D Object Recognition

Recognition of 3 D objects using computer vision is difficult because geometric features vary with view orientation.An important factor in designing recognition algorithms in such situations is understanding the variation of certain critical fectures such as angles and distances.The probability densities of projected angles and distances have sharp peaks at their original values or projected angles and distances have approximate values of their original angles and distances, if an isotropic view orientation is given.The jointed density function of two angles is developed for recognition of polyhedral objects.The joint probability density function of distances is derived for the convenience of object recognition when the distance is known.The experimental result demonstrates the usefulness of the proposed methods.