How to detect wood surface defects?

Research on online detection technology of wood surface defects based on machine vision

The wood surface defects detection technology based on machine vision technology and pattern recognition theory of MOST Technology has the advantages of non-destructive, rapidity, accuracy and economy, which is very important for the sawn timber grade sorting, improving the quality of sawn timber and accelerating the automation of wood processing. In this paper, three common wood defects, such as insect eyes, dead knots and live knots, are taken as the research objects, and the machine vision detection method of wood surface defects is studied in depth. The main content includes: wood surface defect preprocessing, image segmentation, feature extraction, defect type recognition. Image preprocessing is the first step, according to the characteristics of the wood surface image, enhance and sharpen the image to eliminate noise. Image segmentation is a key step in the detection of wood surface defects, for the traditional Ostu algorithm and Renyi entropy algorithm deficiencies, according to the wood defects of this natural texture-type things, the wavelet reconstruction method is used to segment the wood defects image, the method applies the wavelet basis function in the optimal decomposition level of the texture image decomposition, and then in the better resolution level of correctly selecting the smoothing image or details of the image to reconstruct the image, and in the resolution level, the image is reconstructed. In the reconstructed image, the uniform texture pattern is effectively removed and only the local defective regions are retained. Finally, mathematical morphology tools were used to perform morphological post-processing on the segmented images to enhance the visibility and integrity of the segmented images and improve the accuracy of defect extraction. For the recognition of wood defects, the defects were described from two perspectives: texture features (11 gray matrix parameters) and geometric features (elongation and rectangularity). A BP neural network classifier was used to recognize the wood defects, and the recognition rate reached 90%. The experimental results proved that: the use of machine vision technology, according to the image texture characteristics of the defects on the surface of the wood, the defects on the surface of the wood for segmentation and recognition processing is an effective way.

QCROBOT can provide this machine vision module and engineering solutions

.