Picture restoration method and system

Abstract

The invention provides a picture restoration method and system. The method comprises the following steps: acquiring a picture shot by a user and location information of the shot picture; determining the contour of an object in the shot picture through semantic segmentation, and determining a filling area according to the contour of the object; finding out a picture most similar to the shot picturefrom pictures stored in a database as a reference repair picture according to the picture shot by the user and the location information of the shot picture; and through deep learning, filling and repairing the filling area according to the reference repairing picture. According to the method, the problem that the picture is filled with the irrelevant object in the shot picture is solved, so thatthe repaired picture better conforms to the real scene of the shooting place.

Description

technical field [0001] The present invention generally relates to the field of image processing, and more specifically, relates to a method and system for image restoration. Background technique [0002] At present, with the development of social networks and the improvement of mobile phone shooting performance, more and more people choose mobile phones as their shooting tools, and take pictures as souvenirs or share them with friends, but there are often some irrelevant pictures in the pictures. The object affects the effect of the captured picture. For example, when traveling, especially on holidays, various famous tourist attractions are overcrowded, and the pictures taken by users at the scenic spots are full of other tourists, which seriously affects the effect of the pictures taken and the user experience. [0003] The current image repair method requires the user to manually select which areas to fill or repair the background. For example, through the operation previ...

AI Technical Summary

Problems solved by technology

In addition, the current image repair method is relatively simple and rough, without considering the actual shooting location and scene information, but only filling or repairing according to the pixel conditions around the selected filling area, resulting in the final repair effect being unnatural or different from the actual shooting time. The scene information is very different. For example, according to the relative position of the masked image in the preview interface, the same processing is done in the photo image; the photo output image is superimposed to cover or fill the photo image in the same proportion.

Images