A novel image segmentation approach for wood plate surface defect classification through convex optimization

Abstract

Detection of wood plate surface defects using image processing is a complicated problem in the forest industry as the image of the wood surface contains different kinds of defects. In order to obtain complete defect images, we used convex optimization (CO) with different weights as a pretreatment method for smoothing and the Otsu segmentation method to obtain the target defect area images. Structural similarity (SSIM) results between original image and defect image were calculated to evaluate the performance of segmentation with different convex optimization weights. The geometric and intensity features of defects were extracted before constructing a classification and regression tree (CART) classifier. The average accuracy of the classifier is 94.1% with four types of defects on Xylosma congestum wood plate surface: pinhole, crack, live knot and dead knot. Experimental results showed that CO can save the edge of target defects maximally, SSIM can select the appropriate weight for CO, and the CART classifier appears to have the advantages of good adaptability and high classification accuracy.

Keywords

Convex optimization / Threshold segmentation / Structure similarity / Decision tree / Defect recognition

References

[1] Baják S, Páles Z. A separation theorem for nonlinear inverse images of convex sets. Acta Math Hung, 2009, 124(1): 125-144.

[2] Barszcz A, Michalec K, Wasik R. A comparative analysis of pine and beech wood classification according to polish and eunorms. Drewno, 2014, 57(193): 135-145.

[3] Bhateja V, Srivastava A, Kalsi A. Fast SSIM index for color images employing reduced-reference evaluation. Adv Intell Syst Comput, 2014, 247: 451-458.

[4] Bouhamidi A, Enkhbat R, Jbilou K. Conditional gradient Tikhonov method for a convex optimization problem in image restoration. J Comput Appl Math, 2014, 255: 580-592.

[5] Castellani M, Rowlands H. Evolutionary artificial neural network design and training for wood veneer classification. Eng Appl Artif Intell, 2009, 22: 732-741.

[6] Dirami A, Hammouche K, Diaf M, Siarry P. Fast multilevel thresholding for image segmentation through a multiphase level set method. Sig Process, 2013, 7: 342-355.

[7] Han YH, Wang YM, Sun Q, Zhao Y. Crop row detection based on wavelet transformation and Otsu segmentation algorithm. J Electron Inf Technol, 2016, 1: 63-70.

[8] Hu CS, Min X, Yun H, Wang T, Zhang SK. Automatic detection of sound knots and loose knots on sugi using gray level co-occurrence matrix parameters. Ann For Sci, 2011, 68(6): 1077-1083.

[9] Santiago AJS, Yuste AJ, Expósito JEM, Galán SG, Prado RP, Maqueira JM, Bruque S. Real-time image texture analysis in quality management using grid computing: an application to the MDF manufacturing industry. Int J Adv Manuf Technol, 2012, 58(9): 1217-1225.

[10] Wang DQ, Liu ZW, Cong FY. Wood surface quality detection and classification using gray level and texture features. Lect Notes Comput Sci, 2015, V9377: 248-257.

[11] Yuan WQ, Li SL, Li DJ. Wood surface defect recognition based on sub-region zoom Gaussian fitting. Chin J Sci Instrum, 2016, 4: 879-886.

[12] Zhang YZ, Liu SJ, Cao J, Li C, Yu HL. A rapid, automated flaw segmentation method using morphological reconstruction to grade wood flooring. J For Res, 2014, 25(4): 959-964.

[13] Zhang YZ, Liu SJ, Cao J, Li C, Yu HL. Wood board image processing based on dual-tree complex wavelet feature selection and compressed sensing. Wood Sci Technol, 2016, 50(2): 297-311.