Position privacy protection method in social network picture sharing and social network platform

Abstract

The invention belongs to the technical field of visual privacy protection, and discloses a position privacy protection method in social network picture sharing and a social network platform. The method comprises the following steps: carrying out image segmentation and clustering preprocessing on a collected place image set to obtain classification labels of objects in an image, screening the classification labels, and carrying out training of an object detection model; according to the statistical information of various objects in the data set, judging the privacy degree of the objects from the number strategy and the relative position strategy respectively, and training a privacy discrimination model after obtaining a privacy label; carrying out privacy elimination on the privacy region in an image restoration mode according to different degrees; and taking the feedback of a user as data to be input into the privacy discrimination model for updating, so that the standard reliability is ensured. When disturbance to actual image content is very small, a machine target detection method cannot correctly detect a target existing in an image and cannot be associated with actual positioninformation. According to the invention, the position privacy in the image content can be well protected.

Description

technical field [0001] The invention belongs to the technical field of visual privacy protection, and in particular relates to a location privacy protection method and a social network platform in social network picture sharing. Background technique [0002] At present, the closest existing technology is as follows: In recent years, with the rapid development of mobile Internet technology and hardware level, it is easier and easier for people to access and obtain various online services. Among them, various social network platforms with different functions, as the communication carrier of social activities, occupy most of the time in people's daily life, allowing people to share every detail of their lives anytime, anywhere. However, on public social platforms, if there is no suitable processing method, private information is likely to be leaked and abused, causing harm to users themselves, especially the most intuitive picture data, which may be exposed in the public due to...

AI Technical Summary

Problems solved by technology

[0004] To sum up, the problems existing in the existing technology are: for the privacy of social networks, subject to cumbersome and general conditions, means based on access control usually cannot achieve better results; for location privacy, the current Most of the technology is aimed at the actual physical location of the device in the LBS, but there is no method to deal with the location privacy leakage in the image; and for the image analysis method combined with data mining and convolutional neural network, there are few privacy protection methods to counteract it.

Images