BP Neural Network-Based Stripe Width Computation For Adaptive Control Of Line Structured Light Sensors

摘要

To enhance the measurement quality of line structured light sensor (LSLS), an adaptive control method is proposed which can modulate the average stripe width (ASW) within a favorite range by tuning of camera exposure time. As the prerequisite to achieve the control, the ASW is computed based on the back-propagation (BP) neural network that can be well trained in advance. During measurement process, only the forward computation is needed and the computation efficiency can be guaranteed. According to the approximate linear relationship between ASW and exposure time, a linear iteration method is developed to obtain the favorite exposure time. The experiment results show that this method can avoid the over/under exposure of stripe and lead to smoother center extraction results. Moreover, the measurement integrity can also be improved. The proposed method does not need the modification of sensor hardware, and can be easily applied.