基于帧间灰度变化分析的在线光源位置计算

针对机器视觉系统在实际生产中光源位置无法事先确定的问题，提出一种基于帧间特征区域灰度形态分析的在线光源位置计算方法。首先基于灰度分布来确定一特征区域作为参考点，然后运用块匹配算法确定相邻两帧中特征区域的位置变化，之后结合光照模型建立灰度和几何位置之间的关系，运用线性回归对联立方程组进行求解，最后得到光源位置。实验结果表明：光源位置的计算结果与实际所测的距离相比，误差在5%之内。所提算法已被应用于实际生产，具有较好的计算精度和实时性。

On-line light source position calculation based on inter-frame gray-scale variation analysis

Aiming at the problem that the position of the light source cannot be determined in practical production, an on-line position calculation for the light source was proposed, which fully analyzed the gray-scale variation of the feature region in consecutive frames. In this approach, a featured region was first selected based on the local gray-scale distribution and used as a reference point, then its position was tracked in the following consecutive frames via block matching method. Combing with the analytic calculation from the light transport model, the relationship between the gray-scale value and geometric position was established. Finally, the set of equations were solved by linear regression method to find the position of light source. The experiment results show that the errors between the calculated results and the measured results are within 5%. The proposed approach has been successfully applied in the real-time manufacturing with good estimation accuracy.