Pengembangan Teknik Digital Watermarking pada Citra Medis dengan Fitur Pemulihan Gambar

Abstract

Kemajuan teknologi telah mempercepat pemanfaatan gambar digital diberbagai bidang. Namun, kemudahan dalam mendistribusikan gambar digital turutmeningkatkan risiko terhadap berbagai serangan, seperti pencurian kekayaanintelektual dan manipulasi data. Salah satu solusi untuk menjaga keaslian dankepemilikan gambar digital adalah melalui teknik watermarking. Penelitian inimengembangkan metode digital watermarking dengan fitur pemulihan gambaryang diterapkan pada citra medis. Metode yang diusulkan bertujuan untukmemverifikasi integritas data sekaligus memulihkan bagian gambar yangmengalami kerusakan. Implementasi menggunakan algoritma berbasis LeastSignificant Bit (LSB) dan Arnold’s Map menunjukkan hasil yang baik, dengan nilaiPeak Signal-to-Noise Ratio (PSNR) di atas 44 dB dan nilai Structural SimilarityIndex (SSIM) di atas 0,92. Tingkat deteksi watermark mencapai 100%. Selain itu,pengujian juga membuktikan kemampuan algoritma dalam memulihkan gambarsetelah mengalami berbagai jenis serangan, yang ditunjukkan melalui peningkatannilai PSNR dan SSIM setelah proses pemulihan.

Technological advancements have significantly accelerated the adoption ofdigital imaging across various domains. Nevertheless, the ease with which digitalimages can be distributed has concurrently heightened the risk of maliciousactivities, including intellectual property theft and data tampering. Digitalwatermarking has emerged as a viable technique to safeguard the authenticity andownership of digital images. This study proposes a fragile digital watermarkingmethod equipped with an image recovery feature, specifically designed forapplication in medical imaging. The proposed approach aims to ensure dataintegrity verification while enabling the restoration of tampered image regions. Themethod employs a combination of Least Significant Bit (LSB) embedding andArnold’s Transform, yielding promising results. Experimental evaluationsdemonstrate that the method achieves Peak Signal-to-Noise Ratio (PSNR) valuesexceeding 44 dB and Structural Similarity Index Measure (SSIM) values greaterthan 0.92. Moreover, the watermark detection rate consistently reached 100%. Therobustness of the recovery mechanism was further validated through various attackscenarios, as evidenced by improvements in both PSNR and SSIM metricsfollowing the restoration process.