A Practical System to Predict the Absorption Coefficient, Dimension and Reverberation Time of Room using GLCM, DVP and Neural Network

Abstract

Various prediction techniques of reverberation time such as the Sabine and Eyring equations, ray-method, and numerical method require main parameters such as the absorption coefficient and dimensions. Normally, these parameters are obtained from references or/and measurements that necessitate special equipment and skills. On that matter, the authors have proposed a new practical technique to identify the absorption coefficient and dimensions of rooms. The technique comprises Subsystem_1 and Subsystem_2, each of which uses photographic images. Subsystem_1 uses a Gray Level Co-occurrence Matrix (GLCM) and is integrated with a Neural Network (NN) to identify the absorption coefficient of the material. Subsystem_2 uses a Dimension Vision Predictor (DVP) with the author’s “ruler method” to identify the dimensions. Examinations conducted in practical rooms revealed a good correlation coefficient of r ≥ 0.90 for Subsystem_1 and r ≥ 0.99 for Subsystem_2. Finally, the System using NN gave inconsistent results, while FEA revealed consistent results with r ≥ 0.8.