Synthetic acoustic echo-signal intensity distributions for different types of defects

The dataset was composed by carrying out 3000 sound propagation simulations with 4 acoustic emitters generating pulses and 4 sensors measuring acoustic pressure. The input signals are analyzed using the acoustic energy mapping algorithm based on beamforming techniques, that allows to reconstruct the two-dimensional echo-signal intensity map. Each map is segmented using thresholding and connected component search. Object extraction forms the presented dataset. Each object is placed on the 500x500 pixel canvas, so the scale is preserved. The dataset consists of acoustic echo-signal intensity distribution visualizations for five types of objects. Three types are the defects: "cavern", "corrosion" and "crack". One other type is "mixed" which includes the cases of visualization merging between two closely placed defects. The last type is "noise", which includes all other acoustic map objects that do not correspond to defect types or "mixed".