Nonlinear Trajectory Intra Prediction for Image Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In image and video coding, intra prediction methods using straight-line trajectories fail to accurately predict textural information, leading to low compression rates and poor image quality, especially during scene changes or when no good time predictor is available.
Innovation Solution
A method and device for forming prediction values using nonlinear trajectories that are not straight lines, where the second image value is assigned to the third image value along the trajectory with the shortest distance, allowing for weighted averaging of multiple auxiliary trajectories to estimate the prediction value more accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If straight-line trajectories are used for intra prediction, then the prediction method is simple and computationally efficient, but textural information is not accurately predicted resulting in low compression rates and poor image quality
Solution Approach 1:
The patent applies curved trajectories instead of straight lines for intra prediction. The curved trajectory allows the prediction to follow the natural contours and textures in the image, improving prediction accuracy for textural information while maintaining computational efficiency through predefined curvature patterns.
Solution Approach 2:
The patent introduces dynamic trajectory selection where the prediction path adapts to local image characteristics. By selecting from multiple predefined curved trajectory patterns based on the local image content, the method achieves both accuracy for complex textures and simplicity for smooth regions.
2Measurement precision
If curved trajectories are used for intra prediction, then textural information is accurately predicted improving compression rate and image quality, but the prediction method becomes more complex
Solution Approach 1:
The patent changes the parameter of trajectory shape from linear to curved, and introduces trajectory curvature as a controllable parameter. By adjusting the curvature parameter of the trajectory, the method can adapt to different image regions, achieving high prediction accuracy while managing complexity through parameterized curves.
Solution Approach 2:
The patent segments the prediction process into multiple steps: selecting the appropriate curved trajectory pattern, determining trajectory parameters, and performing the prediction. This segmentation makes the complex process more manageable and allows for optimization at each stage.
3Measurement precision
If multiple auxiliary trajectories are used for prediction, then the prediction value is more accurate, but the computational complexity increases
Solution Approach 1:
The patent uses partial action by selecting only the necessary number of auxiliary trajectories based on local image characteristics. Instead of always using multiple trajectories, the method adapts the number of trajectories used, applying more complex multi-trajectory prediction only where needed, thus balancing accuracy and efficiency.
Data Source
AI summary
A method and device form a prediction value. A prediction direction is locally described by nonlinear trajectories. The prediction direction can be used in forming a prediction value to achieve a more accurate prediction determination. The method and device can be used in image compression or image-sequence compression.