A Blockchain and SIFT Based System for Image Copyright Protection

Jianfeng Shi,Dian Yi,Jian Kuang
DOI: https://doi.org/10.1145/3376044.3376051
2019-12-09
Abstract:Internet technology has driven the dissemination revolution of images, but also makes the illegal reproduction and unauthorized use of images extremely rampant. At present, the registration of digital copyright of images should be authorized by authoritative management agencies. There are many problems, such as long audit cycle, no substantive review, difficulty in proof, high cost, centralized storage and so on. Based on blockchain technology and SIFT local feature extraction algorithm, this paper studies and implements a new generation of image digital copyright system. SIFT algorithm is used to extract the invariant features of image such as angle of view, brightness and rotation, etc. which will constitute the local feature vector set of the image, and these local feature vector sets are regarded as the only copyright basis of the image. Even if the image is modified by scaling, rotation and brightness, local features can still be extracted correctly, which can effectively prevent copyright registration of infringing images; Furthermore, IPFS (Inter-Planetary File System) was utilized for distributed storage of images' copyright features; Finally, Hyperledger Fabric and smart contract (chaincode) are used to realize copyright registration, copyright transfer and other functions. The experimental results show that the system has the advantages of automatic similar infringement detection, decentralized storage, tamper-proof and traceability.
What problem does this paper attempt to address?