Video compression optimized for racing drones

Sammanfattning: This thesis is a report on the findings of diifferent video coding techniques and their suitability for a low powered lightweight system mounted on a racing drone. Low latency, high consistency and a robust videostream is of the utmost importance. The literature consists of multiplecomparisons and reports on the effciency for the most commonly used video compression algorithms. These reports and findings are mainly not used on a low latency system but are testing in a laboratory environment with settings unusable for a real-time system. The literature that deals with low latency video streaming and network instability shows that only a limited set of each compression algorithms are available to ensure low complexity and no added delay to the coding process. The findings resulted in that AVC/H.264 was the most suited compression algorithm and more precise the x264 implementation was the most optimized to be able to perform well on the low powered system. To reduce delay each frameneeds to be divided into sub-frames so that encoding and decoding may be done in parallel independently of other sub-parts of the frame. This also improves error propagation when used together with an All-Intra (AI) mode that doesn't utilize any motion prediction techniques.