Refresh-Area Intra Prediction for Error-Resilient Image Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image coding systems face collapse during wireless transmission due to frequent stream errors, especially in low-delay image transmission scenarios where error propagation is high and difficult to manage.
Innovation Solution
The implementation of an image coding apparatus that uses Intra-frame coding for partial areas, with a slice structure and error-tolerant configurations, such as controlling Intra prediction modes and quantization values, to prevent error propagation and maintain image quality. This includes setting up a refresh area within P-pictures for error reset and using slice structures to isolate errors, ensuring that Intra prediction relies only on pixels within the refresh area and employing random prediction modes to minimize step-differences at boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If Intra-frame coding is used forcefully for partial areas to enable low-delay transmission, then transmission delay is reduced, but error propagation increases causing image collapse
Solution Approach 1:
The image is divided into multiple slices, and each slice is further divided into prediction units. This segmentation allows errors to be contained within specific slices and prediction units, preventing error propagation across the entire image. The patent applies this by organizing the image into a hierarchical structure where coding is performed independently on segmented regions.
Solution Approach 2:
Different prediction modes (Intra-frame and Inter-frame) are applied to different regions of the image based on local characteristics. The patent uses motion compensation for regions with high temporal correlation while using Intra-frame prediction for regions where error propagation risk is higher, optimizing both compression efficiency and error resilience locally.
2Reliability
If Intra-frame coding is used to refresh partial areas, then error propagation is reduced, but coding complexity increases
Solution Approach 1:
The prediction mode selection is dynamic rather than static. The patent employs a mode decision mechanism that adaptively chooses between Intra-frame and Inter-frame prediction for each prediction unit based on motion characteristics, regional error probability, and coding efficiency metrics. This dynamic adaptation optimizes the balance between error tolerance and coding complexity.
Solution Approach 2:
The patent changes key coding parameters including prediction mode, slice configuration, and motion compensation settings to optimize error resilience. By adjusting these parameters dynamically based on content characteristics and transmission conditions, the system achieves improved error tolerance without excessive complexity increase.
3Productivity
If motion compensation is used for efficient coding, then compression efficiency is improved, but error propagation increases
Solution Approach 1:
The patent applies motion compensation selectively rather than universally. Motion-compensated prediction is used for prediction units with high temporal correlation and low error risk, while Intra-frame prediction is used for units where error propagation would be problematic. This partial application maintains coding efficiency while limiting error propagation.
Solution Approach 2:
The patent introduces an intermediate decision-making layer (mode selection mechanism) that mediates between motion compensation and Intra-frame prediction. This intermediary evaluates motion characteristics and error risks, then selects the appropriate prediction method, effectively balancing compression efficiency and error resistance.
Data Source
AI summary
A refresh area is taken advantage of when a series of images are coded using Intra-frame coding and Inter-frame coding, a partial area of an image being forcefully Intra-frame coded (Intra MBs) in the refresh area. As the prediction mode for prediction of the Intra MBs inside the refresh area, a prediction mode is selected where the prediction is executable based on only the image that exists inside the refresh area.