Main content area

Acoustic tomography based on hybrid wave propagation model for tree decay detection

Liu, Lei, Li, Guanghui
Computers and electronics in agriculture 2018 v.151 pp. 276-285
Picus, algorithms, image analysis, lakes, models, sounds, tomography, trees, China
Acoustic wave technologies have been used in wood defect detection for many years. Various tomographic imaging algorithms have been proposed to construct two dimensional acoustic velocity maps that serve as the scientific evidence of tree’s internal condition. Many previously proposed algorithms were based on the assumption that acoustic waves propagate along straight rays in the cross sections of trees. To simulate the acoustic wave propagation more realistically, this paper proposed a tomographic imaging method based on a hybrid wave propagation model (HWPM) that includes two velocity inversion phases: straight-ray inversion and curved-ray inversion. The straight-ray inversion was first applied to obtain the preliminary velocity distribution, followed by the curved-ray inversion to refine the velocity distribution. A velocity correction factor was determined for each grid cell following each inversion iteration, and then was used to derive a new velocity value for each corresponding cell. The maximum and minimum velocity constraints were imposed on the newly-computed velocities, and some velocity constraints of grid cells were also imposed to address the nonuniqueness problem commonly existed in velocity inversion and generate a more accurate tomographic image. To evaluate the effectiveness of the hybrid inversion method, we conducted multipath acoustic wave testing on three short log sections (one sound and two with artificial defect) in the laboratory and four live trees in Yangzhou Slender West Lake Park, Jiangsu Province, China. Time-of-flight data was collected by PiCUS Tomograph tool and used to generate tomographic images with both Simultaneous iterative reconstruction technique (SIRT) and the proposed hybrid inversion method to make a comparison. Micro-drilling tests were conducted on the live trees to determine the internal condition and served as the basis for evaluating the tomographic images. Our results indicate that the hybrid wave propagation model is effective in constructing acoustic tomographic images that are reasonably accurate in detecting the internal defects of urban trees.