PRACTICAL SUBSYSTEMS TO IDENTIFY ROOM DIMENSIONS AND MATERIAL SURFACES USING PHOTOGRAPH IMAGES FOR ROOM ACOUSTIC SIMULATIONS

Abstract

A practical technique for simulating room acoustics parameters is proposed. The technique comprises Subsystems 1 and 2, each of which uses photographic images. Subsystem 1 uses a Gray Level Co-occurrence Matrix and a Feed Forward Neural Network to identify material surfaces. Subsystem 2 uses a Dimension Vision Predictor with the author’s “ruler method” to identify the dimensions. Examinations conducted in practical rooms revealed good correlation coefficients of r ≥ 0.90 for Subsystem 1 and r ≥ 0.99 for Subsystem 2. Finally, simulations of reverberation times were conducted using Finite Element Analysis (FEA) with identified parameters. Sufficient agreement was confirmed.