<p>With the rapid evolution of technology, the distribution and sharing of multimedia content like text, images, audio, and video over the Internet are growing significantly. However, the widespread availability of various software and tools has also led to a surge in malicious tampering and unauthorized alterations of multimedia data. As a result, ensuring image integrity, content authentication, and the restoration of original content has become a critical focus area for researchers. Considering the merits of three different techniques namely Discrete Cosine Transform (DCT), Local Binary Pattern (LBP) and Fuzzy Logic (FL), a novel fragile self-embedding watermarking model has been designed for image authentication, tamper localization and restoration. During the embedding phase, the input image is divided into non-overlapping blocks of size (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(3\times 3\)</EquationSource> </InlineEquation>). Watermark such as authentication bits and restoration bits of every block have been produced from 6 Most Significant Bits (MSBs) of every pixel. Depending on the resemblance among neighboring pixels, the blocks are classified into complex block and smooth block using similarity matrix which is computed by fuzzy logic. Authentication bits are generated in smooth block as well as in complex block using Local Binary Pattern (LBP) descriptor as it preserves the local relationship that exists in the cover image. Due to high similarity existence among neighboring pixels, restoration bits are generated using mean in smooth block. In complex block, restoration bits are generated using Discrete Cosine Transform (DCT) as it has excellent energy compaction property. Authentication bits of a block are inserted into Least Significant Bits (LSBs) of the self block for eliminating False Tamper Detection (FTD) problem of prior methods proposed in the past. Restoration bits are inserted into 2LSBs of mapped block. The efficacious of the presented watermarking method has been investigated by applying different attacks on watermarked image. The efficacy of the strategy is assessed based on Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM), Q-index and Bit Error Rate (BER). The potency of the presented method has been evaluated based on the Tamper Detection Rate (TDR) and Tamper Restoration Rate (TRR). The experimental results exhibit the superiority and efficiency of the proposed method at high tampering rate (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(60\%\)</EquationSource> </InlineEquation>). This program offers substantial advantages for both the public and private sectors, particularly in the fields of intellectual property rights and healthcare services.</p>

Self-embedding tampered image localization and restoration scheme exploiting DCT, LBP with fuzzy logic

期刊论文
发表日期：

2025年8月26日

查看全文

Manasi Jana,
Biswapati Jana,
Subhankar Joardar

摘要

With the rapid evolution of technology, the distribution and sharing of multimedia content like text, images, audio, and video over the Internet are growing significantly. However, the widespread availability of various software and tools has also led to a surge in malicious tampering and unauthorized alterations of multimedia data. As a result, ensuring image integrity, content authentication, and the restoration of original content has become a critical focus area for researchers. Considering the merits of three different techniques namely Discrete Cosine Transform (DCT), Local Binary Pattern (LBP) and Fuzzy Logic (FL), a novel fragile self-embedding watermarking model has been designed for image authentication, tamper localization and restoration. During the embedding phase, the input image is divided into non-overlapping blocks of size ( \(3\times 3\) ). Watermark such as authentication bits and restoration bits of every block have been produced from 6 Most Significant Bits (MSBs) of every pixel. Depending on the resemblance among neighboring pixels, the blocks are classified into complex block and smooth block using similarity matrix which is computed by fuzzy logic. Authentication bits are generated in smooth block as well as in complex block using Local Binary Pattern (LBP) descriptor as it preserves the local relationship that exists in the cover image. Due to high similarity existence among neighboring pixels, restoration bits are generated using mean in smooth block. In complex block, restoration bits are generated using Discrete Cosine Transform (DCT) as it has excellent energy compaction property. Authentication bits of a block are inserted into Least Significant Bits (LSBs) of the self block for eliminating False Tamper Detection (FTD) problem of prior methods proposed in the past. Restoration bits are inserted into 2LSBs of mapped block. The efficacious of the presented watermarking method has been investigated by applying different attacks on watermarked image. The efficacy of the strategy is assessed based on Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM), Q-index and Bit Error Rate (BER). The potency of the presented method has been evaluated based on the Tamper Detection Rate (TDR) and Tamper Restoration Rate (TRR). The experimental results exhibit the superiority and efficiency of the proposed method at high tampering rate ( \(60\%\) ). This program offers substantial advantages for both the public and private sectors, particularly in the fields of intellectual property rights and healthcare services.