By the development of technology digital contents have became more and more important in our life. Copyright protection can be supported with automated content recogniser techniques. There are two methods to achive that: feature extraction and watermarking. Watermarking is a popular method, but it has a serious issue: watermark can be removed. A viable alternative can be the feature extraction method to prevent the spread of a copyrighted content on video-sharing websites. These methods are based on computer vision algorithms.
The goal of this thesis is to develop an automatic content recogniser system. First I made some research in this area. I read and processed all the available documents, articles and algorithms. Based on this research I have implemented the existing solutions. The most of the algorithms can be paralellized so my next task was to design the paralellization of the found algorithms. The implementation step was followed by the testing and measuring the algorithm’s robustness.