Scribd
Upload
45 views

A Privacy-Preserving Image Retrieval Scheme

The document proposes a privacy-preserving image retrieval scheme that allows encrypted images to be searched and similar images retrieved. It encrypts images using big-block permutation, pixel permutation, and polyalphabetic cipher to protect content while supporting feature extraction. Local Binary Pattern features are extracted to generate feature vectors for each encrypted image using bag-of-words modeling. Similarity between encrypted images is measured by Manhattan distance of feature vectors, outperforming existing schemes in security and retrieval accuracy.

Uploaded by

Vijay Arapath
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
45 views

A Privacy-Preserving Image Retrieval Scheme

The document proposes a privacy-preserving image retrieval scheme that allows encrypted images to be searched and similar images retrieved. It encrypts images using big-block permutation, pixel permutation, and polyalphabetic cipher to protect content while supporting feature extraction. Local Binary Pattern features are extracted to generate feature vectors for each encrypted image using bag-of-words modeling. Similarity between encrypted images is measured by Manhattan distance of feature vectors, outperforming existing schemes in security and retrieval accuracy.

Uploaded by

Vijay Arapath
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 11

Image Retrieval Scheme Using

Secure Local Binary Pattern


Abstract
 In this project a privacy-preserving image retrieval scheme, in
which the images are encrypted but similar images to a query can
be efficiently retrieved from the encrypted images.

 the image content is protected by big-block permutation, 3 x 3


block permutation within big-blocks, pixel permutation within 3
x 3 blocks, and polyalphabetic cipher.

 The use of polyalphabetic cipher improves security and causes no


degradation in terms of retrieval accuracy as the substitution
tables are generated by the order-preserving encryption.
 The secure local LBP(Local Binary Pattern) features are used
to generate the feature vector for each image by the bag-of-
words model.

 Finally, the similarity among the encrypted images is


measured by the Manhattan distance of such feature vectors.

 The security analysis and experimental results demonstrate


that the proposed scheme outperforms the main existing
schemes in terms of security and retrieval accuracy.
Existing System
 The existing privacy-preserving CBIR methods can be divided
into two categories, i.e., the feature-encryption based and image-
encryption based schemes.

 Feature-encryption based schemes: In this kind of schemes, the


image owners need to extract visual features from the plaintext
images firstly.

 Then, the images are encrypted by standard cryptographic


methods and the extracted features need to be encrypted by
specially-designed methods to support similarity computation.
Drawbacks
 The encryption methods used in this kind of schemes are usually
very complex, such as homomorphic encryption, which is very
computation consuming, and the image owner has to undertake
the heavy calculation task.

 This high computational complexity prohibits such schemes from


being deployed in smart devices with limited computing power.

 In the image-encryption based scheme, the image owners only


take the responsibility of image encryption.
Proposed System
The proposed scheme consists of three core algorithms, including
KeyGen, ImgEnc, and FeaExt.

In the initial phase, the image owner carries out KeyGen and
ImgEnc to encrypt images.

After receiving the encrypted images, the cloud server executes


FeaExt to generate the secure index. In the search phase, the image
owner encrypts the query image by ImgEnc and uploads it.

Once receiving the encrypted query, the cloud server extracts the
query feature vector by FeaExt and search on the index with it.
Proposed(cont…)
 Given this dataset of area rug, carpet, keyboard, and wrapping
paper, our goal is to extract Local Binary Patterns from these
images and apply machine learning to automatically recognize
and categorize these texture images.
Advantages
 An encryption method of images is proposed to protect image content and
meanwhile support the feature extraction for image search.

 In our scheme, the polyalphabetic cipher provides good security but cause no
decrease to retrieval accuracy, no matter how many substitution tables are used.

 The BOW model is used to generate the feature vector with the secure LBP
features.

 It is found that the secure LBP features extracted from fully-permuted


big-blocks are also usable for image retrieval. 
Architecture
Modules
 Key Generation

 Image Encryption

 Feature Extraction

 Similar Image Search

 Image Decryption

 Index Construction and Image Updating

 LBP implementation
Algorithm Used

The proposed scheme consists of three core algorithms, including


 KeyGen
 ImgEnc
 FeaExt
 Local Binary Patterns

You might also like