Restoration algorithm of damaged mark image based on texture synthesis

Image restoration technique is very important in the field of industrial vision system, especially in the application of detection and positioning, but now researches for industrial damaged mark image restoration technology is few, which restricts the efficiency and development of detection and location in industrial production. Aiming at some disadvantages existing in the Criminisi algorithm, we first improve the priority function, which can improve the accuracy of image restoration order; Meanwhile, making full use of the variance to select the size of template adaptively, then filling image accurately; Finally, confidence update way is improved to reduce the matching error. Experiments show that the improved method can obtain better effect in image restoration.