Secure and Recoverable RGB-Colored Two-Dimensional Barcodes: A Hybrid Framework Combining Lightweight Cryptography and Pretrained Vision Models

Two-dimensional (2D) barcodes are now embedded in payment platforms, authentication workflows, industrial traceability, smart packaging, and mobile information services. Their ubiquity has simultaneously increased the incentive for phishing, tampering, and malicious redirection, while recent RGB-colored barcode designs have introduced a second challenge: maintaining reliable payload recovery under non-ideal capture conditions. This study presents a unified framework for secure and recoverable RGB-colored 2D barcodes across QR Code, Data Matrix, Aztec, and PDF417 symbologies. The framework combines channel-separated RGB encoding, lightweight hybrid cryptographic protection, and pretrained vision-based validation to jointly improve confidentiality, authenticity, and operational trust. A recoverability-oriented evaluation protocol is introduced to quantify robustness under distance variation, angular distortion, illumination change, blur, and color shift. Experimental results show that compact schemes based on ChaCha20-Poly1305 and Ed25519 achieve the most favorable trade-off between security overhead and decoding reliability, while EfficientNet-B0 offers the best deployment balance among the evaluated vision backbones. Data Matrix and Aztec exhibit the strongest maximum reliable distance under the tested conditions. The results indicate that secure barcode design cannot be treated as a purely cryptographic or purely visual problem; instead, practical deployment benefits from a layered architecture in which cryptography, computer vision, and recoverability metrics are optimized together.