Adaptive exposure time prediction in hyperspectral bands for industrial cameras

In this study, a new method for exposure time correction for hyperspectral imaging is introduced. Initially, the hardware setup is established. Then, a look-up table which shows the minimum and maximum exposure times for each band is built. The images having different exposure times for different hyperspectral bands are acquired using developed image acquisition system. Afterwards, various features that can represent the exposure state are identified and a dataset is established. The success of Multilayer Perceptron model of Artificial Neural Networks (ANN), Linear Regression and REPTree algorithms in determining the quality of exposure are compared. After that, by using the REPTree algorithm that estimates the exposure quality with 99.18% accuracy, an application attempting to determine the image with the highest exposure quality at the desired hyperspectral band is developed and the results are analyzed.