Study on the method of nonmetallic defects based on ultrasonic tomography and morphology

Ultrasonic tomography is a practical method to detect cracks or voids in nonmetallic materials. The typical inversion results were presented by grids based velocity map, and it can not characterized the rapid changes or discontinuities in defects area. When the grid is dense, the solution is non-unique, and the regularization and stochastic method were induced to solve it. However, the regularization constraint will induce subjective judgment thus distort the results, and the using of stochastic method is highly computational cost. Both of them will inhibit the in-situ application. Here we present a new algorithm to depict the defects boundary. The process base on the rough tomography results, which enhance the solution's uniqueness. The model after interpolation will be refined to expected resolution, and we further perform morphological operation to complete defects identification. Fast marching method is induced to calculate the refine model travel time. The output of the proposed algorithm will directly give the optimal solution of defects, and eliminate the human judgment. Two filed data examples will be shown where the tomography results were compared with the algorithm result. The result shows that the proposed algorithm can recur the defects contour properly, without extra measurements and excessively computational cost.