Image restoration on the defects of knots in the veneer based on image decomposition

This thesis makes an analysis of the problems of the Chinese peeling veneer in the process of veneer production and the importance of the texture in industrial production. In this thesis, the author puts forward the theory of image restoration, and applies this method to the surface of the defects of the knots in the veneer which can be repaired automatically. In this way we can improve the automatic level of digging up intelligence section in the process of peeling veneer production. As for the situation that today´s mainstream image algorithm does not have a good repairing effect on part of the texture as well as on the structure the author proposes a new decomposition-based image restoration scheme. Firstly, we can extract some structural elements and textures based on minimizing the total variation denoising algorithm decomposition. When we restore the structural image, it must be taken into consideration that the BSCB method which is commonly applied to images of the BSCB algorithm can only repair smaller defects; We use the extended sample-based fast restoration Criminisi algorithm to repair the texture part, and finally fix good image synthesis and demonstrate the effectiveness of the method by numerical experiments. The results show that the effect of this method for the restoration of the gray image is better than a single image structure repair and a single texture repair.