Classification Techniques Applied in Wood Logs Tomographic Image

• Antonio Alberto Pereira Junior UNICAMP http://orcid.org/0000-0002-2304-3068
• Marco Antonio Garcia de Carvalho UNICAMP https://orcid.org/0000-0002-1941-6036

DOI: https://doi.org/10.5753/sibgrapi.est.2023.27453

Resumo

The forestry assessment field uses the Non-Destructive Tests (NDTs) to analyze woods logs. To assist the identification of anomalies inside the trunks, the ultrasonic tomography can be used as an alternative. With this technique is possible to evaluate the internal conditions of wooden trunks, through wave propagation applied at specific points. To help the identification and analysis of defects in the tomographic image, this work uses resources from the area of Machine Learning approach in order to identify tomographic images with anomalies. In this study is considered three different classifiers: K-Nearest Neighbors (k-NN), Support Vector Machine (SVM) and Convolutional Neural Network (CNN). The experiments performed were compared using metrics: Accuracy, Precision and Recall. To carry out the experiments, a dataset with 5000 ultrasonic tomography images was build using the Data Augmentation process. In the first experiment, the metrics are calculated based on texture descriptors. The best accuracy results obtained for the CNN, SVM and k-NN models were respectively 89.00%, 80.70% and 79.81%. In the second experiment, the anomaly segmentation was performed with Otsu’s segmentation, then it was tested in the SVM classification model. The results found that the SVM model has superior results of demarcation when compared to the CNN model.

Palavras-chave: Ultrasonic tomography, Non-Destructive Tests, Wood defects

Referências

P. Amaral, A. Veríssimo, P. Barreto, and E. Vidal, “Floresta para sempre: um manual para a produção de madeira na amazônia,” WWF, Brasilia, DF (Brasil) Instituto do Homem e Meio Ambiente da Amazonia, Tech. Rep., 1998.

V. Bucur, Acoustics of wood. Springer Science & Business Media, 2006.

S. Brazolin, “Biodeterioração, anatomia do lenho e análise de risco de queda de árvores de tipuana, tipuana tipu (benth.) o. kuntze, nos passeios públicos da cidade de são paulo, sp,” 2009.

C. J. Lin, Y. C. Kao, T. T. Lin, M. J. Tsai, S. Y. Wang, L. D. Lin, Y. N. Wang, and M. H. Chan, “Application of an ultrasonic tomographic technique for detecting defects in standing trees,” International Biodeterioration and Biodegradation, vol. 62, no. 4, pp. 434–441, 2008.

X. Du, J. Li, H. Feng, and H. Hu, “Stress wave tomography of wood internal defects based on deep learning and contour constraint under sparse sampling,” in Intelligence Science and Big Data Engineering. Big Data and Machine Learning, Z. Cui, J. Pan, S. Zhang, L. Xiao, and J. Yang, Eds. Cham: Springer International Publishing, 2019, pp. 335–346.

S. Palma, R. Goncalves, A. Trinca, C. Costa, M. Reis, and G. Martins, “Interference from knots, wave propagation direction, and effect of juvenile and reaction wood on velocities in ultrasound tomography,” BioResources, vol. 13, no. 2, pp. 2834–2845, 2018.

L. V. Socco, L. Sambuelli, R. Martinis, E. Comino, and G. Nicolotti, “Feasibility of ultrasonic tomography for nondestructive testing of decay on living trees,” RESEARCH IN NONDESTRUCTIVE EVALUATION, vol. 15, pp. 31–54, 01 2004.

V. Bucur, “Ultrasonic techniques for nondestructive testing of standing trees,” Ultrasonics, vol. 43, no. 4, pp. 237 – 239, 2005. [Online]. Available: [link].

L. Zeng, J. Lin, J. Hua, and W. Shi, “Interference resisting design for guided wave tomography,” Smart Materials and Structures, vol. 22, no. 5, p. 055017, 2013. [Online]. Available: [link].

X. Du, S. Li, G. Li, H. Feng, and S. Chen, “Stress wave tomography of wood internal defects using ellipse-based spatial interpolation and velocity compensation,” BioResources, vol. 10, no. 3, pp. 3948–3962, 2015.

H. Feng, Z. Qian, M. Hu, Z. Zheng, and X. Du, “The study of stress wave tomography algorithm for internal defects in rl plane of wood,” in 2018 Chinese Automation Congress (CAC), 2018, pp. 2283–2288.

J. Strobel, M. Carvalho, R. Gonçalves, C. Pedroso, M. Reis, and P. Martins, “Quantitative image analysis of acoustic tomography in woods,” European Journal of Wood and Wood Products, vol. 76, 06 2018.

X. Du, J. Li, H. Feng, and S. Chen, “Image reconstruction of internal defects in wood based on segmented propagation rays of stress waves,” Applied Sciences, vol. 8, p. 1778, 09 2018.

X.-d. Zhu, J. Cao, F.-H. Wang, J.-p. Sun, and Y. Liu, “Wood nondestructive test based on artificial neural network,” in 2009 International Conference on Computational Intelligence and Software Engineering. IEEE, 2009, pp. 1–4.

H. Mu, M. Zhang, D. Qi, S. Guan, and H. Ni, “Wood defects recognition based on fuzzy bp neural network,” Int. J. Smart Home, vol. 9, pp. 143–152, 2015.

M. R. Effendi, R. Willyantho, and A. Munir, “Back propagation technique for image reconstruction of microwave tomography,” in Proceedings of the 2019 9th International Conference on Biomedical Engineering and Technology, 2019, pp. 186–189.

M. Hansson, A. Enescu, and S. S. Brandt, “Knot detection in x-ray images of wood planks using dictionary learning,” in 2015 14th IAPR International Conference on Machine Vision Applications (MVA). IEEE, 2015, pp. 497–500.

J. R. A. Strobel et al., “Método de interpolação baseado em elipses associado à análise contextual de rotas para geração de tomografias ultrassônicas em toras de madeira,” 2017.

A. A. P. Junior and M. A. G. de Carvalho, “An initial study in wood tomographic image classification using the svm and cnn techniques.” in VISIGRAPP (4: VISAPP), 2022, pp. 575–581.

C. Shorten and T. M. Khoshgoftaar, “A survey on image data augmentation for deep learning,” Journal of Big Data, vol. 6, no. 1, pp. 1–48, 2019.

N. Otsu, “A threshold selection method from gray-level histograms,” IEEE transactions on systems, man, and cybernetics, vol. 9, no. 1, pp. 62–66, 1979.