One depth camera, running at 90 frames per second can generate millions of depth pixel data points per second. With such a large amount of data, there’s an obvious need for compression to assist with data storage and transmission, especially crucial when you move from single camera usages to multiple camera setups, like those used in volumetric capture.

There are already a variety of methods developed by researchers worldwide for compressing depth images, however many of these algorithms rely on entirely novel and unique approaches which require custom proprietary software, and cannot take advantage of existing hardware acceleration blocks that already exist on many compute platforms.

The methods we are sharing today involve colorizing the depth images, and then using existing RGB tools to compress, store and decompress the data. By utilizing this approach, we can capitalize on existing advanced RGB techniques, while successfully retaining the quality depth data that RealSense Depth cameras can output.

This whitepaper covers the colorization and recovery methods, an application example of compression using a lossy image codec, and the considerations necessary to ensure optimal results.

Read the whitepaper here.