A hidden object detection method based on a random reversible matrix

Abstract

The invention provides a hidden object detection method based on a random reversible matrix, which can be carried out in a safe mode, and meanwhile, protects the privacy of a user picture and the privacy of server algorithm parameters. The cloud server stores various algorithm parameters of object detection, such as a face detection algorithm, and the client hides the image pixels of each region of the image sliding by the convolution kernel by using the matrix, so that the cloud cannot calculate the image pixels and cannot restore the original image. A reversible matrix is introduced into theclient, so that the privacy of the client image is effectively protected. The reversible random matrix is applied to the object detection security protocol for the first time. Any encryption algorithm is not introduced, and the efficiency of engineering application is greatly improved on the premise that the data security of the two parties is ensured. And the method is very easy to realize through software, and can be widely applied and popularized to the technologies of cloud computing, secure image convolution and the like.

Description

technical field [0001] The invention belongs to the fields of cryptography and computer vision, in particular to a method for detecting a hidden object, in particular to a method for detecting a hidden object based on a random reversible matrix. Background technique [0002] Face detection refers to determining the position and size of all faces (if present) in the input image. The input of the face detection system is an image that may contain a face, and the output is a parametric description of whether there is a face in the image, the number, position, and scale of the face. At present, there are many face detection algorithm models, such as ANN model, SVM model, Adaboost model, etc. The Viola&Jones face detection algorithm uses the Adaboost model, which is optimal in terms of speed, robustness and precision. Therefore, the face detection algorithm used in this paper is the Viola&Jones face detection algorithm. [0003] With the proliferation of surveillance cameras, c...

AI Technical Summary

Problems solved by technology

The main problem of the Secure Classifier protocol is a large number of encryption and decryption calculations and the lack of use of integral images to accelerate face detection, which seriously affects the calculation speed of face image stealth detection algorithms based on inadvertent transmission.

Experiments show that it takes a few minutes for a 24×24 detection window to be detected, and a 240×320 picture has about 150,000 detection windows, so it takes a lot of time to detect a 240×320 picture, so it takes a lot of time to detect a picture cost too much

In summary, the stealth detection algorithm for face images based on inadvertent transmission is not very ideal

Images