基于图像融合的木板表面缺陷检测研究

针对木板表面节子缺陷被染料染色后难以识别的问题，本文利用图像融合技术提出一种新的木板表面缺陷检测方法。该方法采集被染色木板的近红外图像和可见光图像，使用加权平均法、主成分分析（PCA）算法、小波变换、Laplacian金字塔变换等不同的融合算法对采集的近红外图像和可见光图像进行融合，然后对不同算法融合后的图像仔细观察分析和比对并计算其信息熵。实验结果证实，融合后的图像能够明显地辨别出染色后的木板缺陷，并且基于Laplacian金字塔算法的融合效果最好。

Wood surface defect detection based on image fusion

This paper proposes a novel wood surface defect detection scheme based on image fusion to solve the detection of wood surface knots stained by colorants. This scheme picks up the infrared and visible images of the stained wood veneer. The weighted average scheme, PCA scheme, wavelet transform scheme and Laplacian pyramid scheme are used to fuse the original images. The fused images with different fusion schemes are compared and investigated carefully. Experiments indicate that fused images can identify the stained knots clearly, and the best outcomes are achieved by the Laplacian pyramid scheme.