A non-invasive method for the cataloguing and authentication of surveillance video using on-camera blockchain participation, machine learning and signal analysis

Presented here is a non-invasive system capable of cataloguing, identifying, and authenticating surveillance video throughout its lifecycle, with particular application to law enforcement. It delivers benefits similar to digital watermarking by utilising signal processing, machine learning object detection and a supporting blockchain infrastructure, and is proposed as a method for managing large volumes of data whilst overcoming issues associated with real time application of digital watermarking or hashing. A functional reference system is tested for effectiveness and its performance is analysed on test data, including evaluation of four search optimisation strategies, with a clear performance ranking shown. The system allows for the ability to identify and evaluate video integrity without requiring modification or access to the original video, making it of specific interest to law enforcement applications.