Camera data compression with low latency

February 02, 2018 // By Christoph Hammerschmidt
As part of the development towards autonomous driving, cars are being equipped with more and more cameras. However, this flood of data pushes the internal networks to their limits. Special compression techniques reduce the amount of video data, but provide high latency for encoding. Fraunhofer researchers have adapted video compression in such a way that latency is almost imperceptible. This makes it relevant for use in automated driving.

Today, up to 12 cameras are installed in new vehicle models, mostly in the front or rear lights and side mirrors. An on-board computer installed in the car uses the data for ADAS functions such as lane assistant, parking aid or to identify other road users or obstacles. Benno Stabernack from the Fraunhofer Institute for Telecommunications, Heinrich Hertz Institute, HHI in Berlin predicts that if autonomous driving is being introduced as fast as currently predicted, the number of cameras will continue to increase.

This translates into even more load for the internal data networks of the vehicles. Currently, these networks can process a data volume of about one gigabit per second. With images in HD quality, this amount of data is already achieved with one camera. "The remedy here is compression," says Stabernack. The Fraunhofer HHI earlier has made contributions to the development of the video coding standards H. 264/Advanced Video Coding (AVC) and H. 265/MPEG High Efficiency Video Coding (HEVC). "With these methods, the amount of data can be greatly reduced. This means that more than ten times the amount of data can be transferred," Stabernack emphasizes.