Integrating chaotic encryption with a quantum-inspired nonlinear feature mapping for privacy-aware image intelligence

Recent advances in intelligent visual systems have created a growing need for frameworks capable of simultaneously protecting sensitive image content and supporting reliable machine-based interpretation. Here, we present a unified privacy-aware visual intelligence framework that integrates chaotic image encryption, statistical security assessment, hybrid feature extraction, and a classical nonlinear feature mapping inspired by quantum feature-space representations. The encryption stage employs pseudo-random key generation, permutation–diffusion dynamics, and dual-key XOR transformations to disrupt spatial structure and obscure the visual content of plaintext images. Statistical analyses demonstrate strong randomness and differential-sensitivity characteristics, achieving a mean NPCR of 99.8709%, a mean UACI of 33.4629%, and a mean ciphertext entropy of 7.9972 bits. In the learning branch, deep and handcrafted features are combined and transformed using a quantum-inspired nonlinear mapping before classification. Using the selected Proposed-QINNFM-SVM model, the independent held-out test achieved 100.0000% accuracy, 1.0000 Macro-F1, and 100.0000% balanced accuracy, while repeated stratified five-fold cross-validation across the five evaluated models on the training data identified Proposed-QINNFM-SVM as the strongest repeated-validation configuration, with 91.03 ± 0.20% accuracy, 0.8894 ± 0.0026 Macro-F1, and 87.10 ± 0.24% balanced accuracy. These results are evaluated separately to distinguish independent-test performance from variability across training-data partitions. The proposed framework provides an integrated computational architecture for privacy-aware image protection and downstream visual classification, with potential applications in medical imaging and other sensitive visual-analysis scenarios.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-17
DOI
https://doi.org/10.1038/s41598-026-80946-8
Primary Topic
Chaos-based Image/Signal Encryption
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Integrating chaotic encryption with a quantum-inspired nonlinear feature mapping for privacy-aware image intelligence

Abdolsamad Hamidi, Ali Farmani, Keyvan Barati
4 links
Scientific Reports
Chaos-based Image/Signal Encryption
article

Integrating chaotic encryption with a quantum-inspired nonlinear feature mapping for privacy-aware image intelligence

Abdolsamad Hamidi, Ali Farmani, Keyvan Barati
article en

Abstract

Recent advances in intelligent visual systems have created a growing need for frameworks capable of simultaneously protecting sensitive image content and supporting reliable machine-based interpretation. Here, we present a unified privacy-aware visual intelligence framework that integrates chaotic image encryption, statistical security assessment, hybrid feature extraction, and a classical nonlinear feature mapping inspired by quantum feature-space representations. The encryption stage employs pseudo-random key generation, permutation–diffusion dynamics, and dual-key XOR transformations to disrupt spatial structure and obscure the visual content of plaintext images. Statistical analyses demonstrate strong randomness and differential-sensitivity characteristics, achieving a mean NPCR of 99.8709%, a mean UACI of 33.4629%, and a mean ciphertext entropy of 7.9972 bits. In the learning branch, deep and handcrafted features are combined and transformed using a quantum-inspired nonlinear mapping before classification. Using the selected Proposed-QINNFM-SVM model, the independent held-out test achieved 100.0000% accuracy, 1.0000 Macro-F1, and 100.0000% balanced accuracy, while repeated stratified five-fold cross-validation across the five evaluated models on the training data identified Proposed-QINNFM-SVM as the strongest repeated-validation configuration, with 91.03 ± 0.20% accuracy, 0.8894 ± 0.0026 Macro-F1, and 87.10 ± 0.24% balanced accuracy. These results are evaluated separately to distinguish independent-test performance from variability across training-data partitions. The proposed framework provides an integrated computational architecture for privacy-aware image protection and downstream visual classification, with potential applications in medical imaging and other sensitive visual-analysis scenarios.

Scientific Reports
Lorestan University (IR)
Chaos-based Image/Signal Encryption
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon