Image Coding TRIZ Cases
- Geometry Point Cloud Coding With Logarithmic Coefficients
Logarithmic transform coefficients and zero-run loops simplify G-PCC syntax while reducing memory use and processing complexity.
- Point Cloud Depth Prediction with Adaptive Lists to Reduce Outliers
Adaptive prediction lists reduce point-cloud depth outliers from scenario discontinuity.
- Neighbor-Based Point Cloud Coding With Geometry Model Conversion
Converted neighbor geometry improves point cloud prediction and reduces coding bits.
- Bit Plane Encoding of Data Arrays for Fixed-Rate Compression
Recursive bit-plane encoding reduces storage and bandwidth for image data.
- 3D Point Cloud Motion Compensation for Efficient Encoding
Motion compensation aligns point clouds before encoding to reduce residual data.
- Positional Local Self-Attention Layers for Efficient Vision Models
A memory block of neighboring features enables positional local attention with fewer parameters and improved vision-model accuracy.
- Point Cloud Transmission with Octree Encoding and Attribute Prediction
This engineering case layers point clouds and uses octree geometry coding to reduce transmission volume, encoding complexity, and latency.
- Neural Network Intra Prediction Transform Adaptation for Video Coding
Neural network intra prediction selects or infers matching transforms, improving compression efficiency while reducing decoding overhead.
- Three-Dimensional Point Cloud Subspace Encoding for Lower Decoding Load
This case uses subspace identifiers in shared control information and data headers to reduce decoding operations for 3D point clouds.
- Visual Item Identification with Pre-Trained Feature Encodings
Pre-trained item and shape classifiers create compact feature vectors for fast matching, enabling new items without retraining.
- Point Cloud Prediction Units for Lower-Latency Transmission
Prediction units and selective motion vectors improve point cloud compression for efficient LiDAR data transmission.
- Triangle Mesh Encoding With Adaptive Angles to Reduce Code Size
This case adaptively encodes shared-triangle vertices and angle prediction errors to reduce 3D mesh code size.
- 2D Data Encoding With 3D Spatial Metadata for Direct Rendering
This case embeds patch separation and 3D spatial metadata in 2D video streams, enabling real-time rendering and patch merging.
- 3D Point Cloud Attribute Encoding for Type-Aware Decoding
Type information and identification fields help decode 3D point attributes accurately while reducing transmitted data volume.
- Video Compression Transformer for Bias-Free Temporal Entropy Coding
This case replaces motion prediction and warping with temporal transformer entropy modeling, reducing complexity for diverse video content.
- Neural Network Feature Tensor Compression for Video Redundancy
Adaptive quantization and entropy encoding target spatial-temporal and inter-view redundancy while preserving coding efficiency.
- Normal-Aware Mesh Decoding for Efficient Face Subdivision
The decoder calculates vertex normals, moves face vertices, and refines subdivisions to reduce bit usage for efficient mesh encoding.
- Mesh Decoding Motion Prediction to Reduce Information Volume
Decoded neighboring vertices, prediction modes, and motion residuals improve mesh encoding efficiency while reducing data volume.
- Mesh Decoding With Base-Mesh Intra Prediction for Efficient Encoding
This mesh decoding approach predicts subdivided vertices from a base mesh, reducing residual information for more efficient encoding.
- 3D Mesh Transmission Using Patch Segmentation and 2D Projection
Preprocessing, patch segmentation, and 2D projection reduce mesh transmission resources while preserving frame correlation for 3D services.