Surveillance is being pervasively used, and its recording is extensively applied in practice. Thereby, protection of surveillance recordings, which is related to multimedia security, increasingly attracts research interests. Traditional techniques, such as watermarking, cryptography and steganography, focus on analysing multimedia content, resulting in high complexity and long latency, which makes them not suitable for protecting real-time surveillance applications. In this paper, we propose a novel blockchain-assisted framework to protect the recordings in real-time surveillance applications. In the proposed framework, we design an algorithm that generates frame fingerprints, which involves real-time extraction of packets from surveillance multimedia streams, SHA3-512 functions, and the verification of frame consistency. We use a blockchain for preserving the frame fingerprints generated by our proposed algorithm. Different from existing works that use simulations to show the performance, we develop a real system using a service-oriented blockchain, virtualisation for distributed ledger technology (vDLT), and the effectiveness of this system is demonstrated.