【摘要】 针对SGM(semi-global matching)算法的惩罚因子在不同场景需要根据经验进行反复测试再进行设定以及惩罚因子为定值，容易出现误匹配的问题，提出SGM算法的惩罚因子自适应动态调整策略，采用惩罚因子根据初始匹配代价与相邻像素点灰度值的梯度在每个像素的各个视差空间下进行自适应动态调整。实验结果表明，在不同的场景下，通过惩罚因子的自适应调节，不仅能够避免设定惩罚因子的难题，还能有效地提高视差匹配的正确率，降低视差图灰度值平均误差。更多还原

【Abstract】 Aiming at the problem that the penalty factor of the SGM algorithm requires repeated testing based on experience for setting in different scenarios, and the fixed value of the penalty factor is prone to mismatches, an adaptive dynamic adjustment strategy for the penalty factor of the SGM algorithm is proposed. The penalty factor is adaptively and dynamically adjusted in each disparity space of every pixel based on the initial matching cost and the gradient of the gray values of adjacent pixels. Experimental results show that in different scenarios, the adaptive adjustment of the penalty factor not only avoids the difficulty of setting the penalty factor, but also effectively improves the accuracy of disparity matching and reduces the average gray value error of the disparity map.更多还原