Multi-scale image restoring algorithm based on structure information

Abstract

The invention discloses a multi-scale image restoring algorithm based on structure information. The multi-scale image restoring algorithm comprises the steps of performing multi-scale decomposition on a to-be-restored image I, and obtaining a series of different scale layer images; from a largest scale layer image, calculating priority of a damaged area of a current k-th scale layer image Ik by means of a priority function calculation method based on the structure information, searching an optimal matching block according to a priority sequence by means of a similarity measuring rule which is composed of corresponding matching rules, performing image restoration of the current k-th scale layer, obtaining a corresponding restoring effect graph and performing up-sampling reconstruction on the restoring effect graph for obtaining a restored comparison reference graph of a (k-1)th scale layer, restoring the image of the (k-1)th scale layer for obtaining a preliminary restoring result in this scale; performing comparison between the preliminarily restoring result in the previous step and the comparison reference graph, and obtaining a new (k-1)-th scale layer damaged image; and obtaining a final restoring effect graph of the (k-1)th scale layer. The multi-scale image restoring algorithm realizes high restoring sequence accuracy, and furthermore realizes more natural and reasonable image structure texture after restoring.

Description

technical field

[0001] The invention belongs to the field of computer image restoration, and relates to a multi-scale image restoration algorithm based on structural information, which can be used to restore deteriorated and deteriorated films, eliminate scratches and the like. Background technique

[0002] Digital image restoration technology is the process of filling information in the missing information area of ​​the image to be repaired. The restorer needs to use the most appropriate method to restore the damaged image, and at the same time make the observer unable to detect the traces of the image being repaired, so as to ensure that the image reaches the best quality. Ideal for artistic effects.

[0003] The most classic digital image restoration algorithm is the restoration algorithm based on sample blocks proposed by Criminisi et al. It uses the priority to determine the sequence of repairs. where the priority is the product of the confidence term and the data ter...

Images