Encoding method and image encoding device for compressing an image sequence

Simple SummaryContent extracted from patent full text and abstract with AI.

This patent discloses a new method and device for compressing image sequences—such as video or medical imaging data—by determining dense motion vector fields and corresponding confidence values for each pixel or voxel between consecutive images. By selecting only the most reliable motion vectors (using confidence values) and transmitting parameters for reconstructing the motion, the approach achieves efficient, high-quality compression, particularly well-suited to complex, deformable motions common in medical imaging, while avoiding common artifacts of block-based methods.

Use CasesContent extracted from patent full text and abstract with AI.

• Efficient storage and transmission of medical imaging sequences (e.g., cardiac CT or MRI scans) where tissue movement is complex and non-rigid.
• High-quality video compression in broadcasts, streaming, or video conferencing, particularly for content involving complicated motion or deformations.
• Compression of industrial or scientific imaging sequences (e.g., non-destructive testing, microscopy, or dynamic tomography) where saving storage bandwidth is crucial.
• Enhancement of archival and telemedicine systems by reducing data size without loss of diagnostically relevant detail.
• Real-time transmission of dynamic 2D or 3D image sequences in telehealth or remote diagnostics.

BenefitsContent extracted from patent full text and abstract with AI.

• Offers significantly higher image sequence compression, even for complex and deformable motion, compared to conventional block-based techniques.
• Reduces storage and transmission requirements for large image datasets, supporting faster communication and lower hardware costs.
• Minimizes prediction errors and avoids visual artifacts (such as blocking) found in traditional compression methods, preserving image quality.
• Enables both lossless and lossy compression to flexibly match legal, diagnostic, or application requirements.
• Can be implemented as software, making it easily deployable or upgradable on existing image processing systems.
• Optimizes the use of only the most reliable motion information, increasing compression efficiency without sacrificing image fidelity.

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

The method (100) involves determining a dense motion vector field (Vn) for an actual image area (In) of an image sequence by comparison with an image area (In-1) of the image sequence. A confidence vector field (Kn) is determined for the actual image area, which indicates a confidence value for a motion vector for the motion vector field. Motion vector field reconstruction parameters (FP) for the actual image area are determined based on the motion and confidence vector fields. The motion vector field is reconstructed by using the reconstruction parameters. Independent claims are also included for the following: (1) a method for decoding an image sequence (2) a method for transmission and/or storage of an image sequence (3) an image coding device (4) an image decoding device (5) a system for transmission and/or storage of an image sequence (6) a computer program comprising instructions for executing a method for compression of image sequence.