The combination of gray level co-occurrence matrix and back propagation neural network for classifying stairs descent and floor

Abstract Smart wheelchairs (SW) technology is one of the solutions to help disabled people who do not have a hand or people to help them. Apart from able to move on its own, a smart wheelchair needs to be safe to use. One of the ways to increase SW safety is the ability to detect obstacles. In this study, we tried to create obstacle detection that can classify the stairs descent and floor based on image processing. To achieve our purpose, Contrast Limited Adaptive Histogram Equalization (CLAHE) is used to increase image contrast. After that, Gray Level Co-occurrence Matrix (GLCM) is used to extract features from the image. Finally, Back Propagation Neural Network (BPNN) is used to classify the image. Based on the test result, BPNN achieves results with 95% Accuracy, 95% Sensitivity, 95% Specificity with an average computation time of 0.0035 s.