A wood defect image detection program was developed using Visual C++ 2017（VC++） and Open CV to detect dead knots and cracks in veneers. First, RGB images were transformed into gray images with fractional-order principal component analysis（PCA）. Then, three channel gray images were segmented with the Chan-Vese（CV） model. Finally, binary images were filtered according to a length-width ratio to detect the dead knots and cracks. The experimental results showed that the algorithm could be used to remove most noise to extract dead knots and cracks in veneers.