High-precision Imaging Algorithm Of Wood Internal Defects Based On Stress Wave

Stress wave nondestructive testing have been widely used in forest and wood defect detection.Compared to other techniques such as X-ray and ultrasonic wave,stress wave testing technique is safer,more portable and applicable,so it has become a popular method for nondestructive detection.With stress wave imaging,the internal defects of trees can be reconstructed.However,the limited number of deployed sensors degrades the accuracy of traditional stress wave imaging methods,which results in the issues such as unrealistic area and location deviation of the internal defects.To this end,this paper proposes three stress wave tomography imaging methods to detect the internal defects of trees,adopting ray segmentation,deep learning,and spatial interpolation into traditional stress wave-based imaging methods.The main contribution of this paper is highlighted as follows:1.To improve the accuracy of stress wave-based tomography imaging,this paper proposes a ray segmentation-based method termed RSIA for it.First of all,RSIA rectifies the stress wave velocity and meshes the ray images generated from stress wave diffusions.Then,every single ray is segmented according to its intersections with other rays,and the velocity of each ray segment is rectified to obtain a more accurate velocity distribution based on the initial speed of stress wave diffusion.To the end,with image processing techniques,the accuracy of tomography imaging is accordingly improved.Experimental results on four logs and four living trees demonstrate that RSIA has better performance than the counterpart methods with respect to the detection accuracy,detected defect area and size.2.Due to the insufficiency of stress wave data,the accuracy of tomography imaging may be degraded.This paper proposes a stress wave tomography imaging algorithm(FRRS)using Fast-RCNN and ray segmentation.The algorithm first establishes a sample library,and then uses the Fast-RCNN model to learn the distribution of stress wave with 6,8,10,and 12 sensors respectively,and obtain the partial priori knowledge of defect samples.Finally,the stress wave tomography imaging with sparse samples is implemented with RSIA algorithm.Experiments on four logs show that FRRS outperforms RSIA and FAKOPP under the condition of sparse samples.3.To reconstruct the 3D image for the internal defects of trees,this paper proposes a stress wave-based 3D reconstruction algorithm based on spatial interpolation termed TDRA.Stress waves velocities in multiple cross sections with different heights are collected,and then the RSIA algorithm is utilized to generate tomographic images as the input of the 3D tomographic imaging method.Then,the linear interpolation method is used to complete the velocity for vacant grids,and all the velocities in grid are updated with the 3D stress wave velocity modification strategy.Finally,color filling is finished according to the velocities of grid units to reconstruct 3D images.Experimental results show that TDRA is feasible for 3D imaging of the internal defects of trees.