A fundamental change in video processing

EuclidVision and the Future of Video Compression: Smart Model Libraries in the Cloud.


A platform technology for the future of video compression.

EuclidIQ’s approach to future video processing represents a fundamental shift in video technology. By extending beyond block-based processing into object and feature modeling, Euclidvision is able to deliver even greater benefits than those available for standards-based encoders. Euclidvision is a suite of tools, algorithms and processing pipelines implementing model-based compression.

EuclidVision extends the block-based compression paradigm to take advantage of features and objects in the video. These features and objects provide a foothold for engaging more advanced modeling, bridging the blocks of conventional compression to feature modeling and object modeling techniques.

When EuclidVision determines that the complexity of the video has exceeded the conventional encoder’s capabilities, it applies higher level modeling to those regions. The higher level of modeling is based on features and objects that are tracked throughout the video.

EuclidVision is able to model the data at multiple levels or fidelities during encoding as shown in the chart.


EuclidVision implementation

EuclidVision goes beyond conventional H.264 and H.265 techniques using model-based compression. Our algorithm detects and identifies features and objects with common characteristics to process video.  The selected features and objects are separated out as Smart Models, which can be stored in libraries and reused to further reduce bandwidth. EuclidVision provides two levels of processing, EuclidVision Compliant (EVC) and EuclidVision Premium, (EVP).

Our EuclidVision Compliant (EVC) implementation resides on the second level or “Block Modeling” level in order to be in conformance with the MPEG standards specifications. Our EuclidVision Premium (EVP) prototype moves up and down the levels as deemed necessary during the encoding process. This feature and object-based compression framework allows EuclidVision to increase modeling complexity as needed to produce the best compression gain.

EuclidVision Compliant (EVC) provides the following:

  • Feature/object detection
  • Block/feature/object tracking
  • Linking to nearby macroblocks
  • Model-based MB predictions
  • Rate-distortion optimization
  • Full compliance with standard decoders

EuclidVision Premium (EVP) adds the following features/functionality in addition to the EVC implementation:

  • Extended reference frames
  • Composite predictions
  • Modified encoding context
  • Modified decoder
  • Smart Model library
  • Offline prediction using Smart Model library

EuclidIQ: The Future of Model-Based Compression

Watch this video to see how a cloud-based architecture of Smart Model Libraries will fundamentally change how video will be processed in the future.