Image Coding TRIZ Cases

Inverting pixel-domain processing by filtering transform coefficients minimizes distortion while lowering encoding computational complexity.

Deriving bilateral filter weights from quantized sample intensity differences reduces computational complexity while maintaining high video quality.

Pointwise prediction derives current points from previously decoded clouds, eliminating spatial and temporal redundancy in LiDAR point cloud coding.

A method adjusts global motion matrix application based on local motion compression presence.

Segmenting rays at voxel boundaries enables parallel intersection testing, reducing computational complexity in direct volume rendering.

Dynamic frame assignment to idle entropy encoders prevents boundary degradation while maintaining maximum throughput.

A polygon rendering method segments arbitrary contours into convex sub-polygons to generate equilateral triangles for efficient hardware processing.

Analyzing GPU counters via machine learning detects scene changes, reducing bandwidth consumption while maintaining visual quality.

Boundary control flags enable cross-slice filter processing using adjacent pixel data, suppressing quality deterioration at slice boundaries during encoding.

Uses luma sample mapping tables to derive chroma intra prediction modes, reducing data transfer costs while maintaining high-resolution image quality.

Segmenting predictor index parsing from attribute reconstruction reduces computational complexity while maintaining parsing accuracy.

Iterative training balances genericity and specialization in neural network intra prediction to improve rate-distortion performance.

Adaptive triangulation compresses small images by mapping pixels to grid points and assigning colors via a table.

A shape-adaptive model-based codec separates binary alpha planes into deterministic and stochastic components to enable flexible lossy and lossless coding.

A G-PCC coder derives coding contexts from reference block planar information to adapt syntax element encoding for current nodes.

Multi-mode feature vectors combine global and text features to enhance image description accuracy without increasing decoder complexity.

A processing system maps a two-dimensional image onto a three-dimensional mesh using depth information to generate stereoscopic views.

Generating depth images from 3D models reduces distributed data volume and network bandwidth usage.

Shared encoding patterns across N-ary and octree structures lower processing load while maintaining coding efficiency.