Video Encoder Area Splitting for Packet Loss Robustness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face challenges in optimizing video encoding, particularly in reducing packet loss errors and improving image quality while maintaining low transmission delay, especially in inter prediction and loop filtering processes.
Innovation Solution
The proposed solution involves an encoder and decoder that split pictures into areas for intra encoding and inter prediction, using only pixels from intra-encoded areas or preceding areas in coding order for reference, and employing adaptive filtering and motion vector correction based on gradient information to minimize packet loss errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If inter prediction is used to improve encoding efficiency and reduce transmission delay, then productivity is improved, but reliability deteriorates due to packet loss errors propagating through reference pixels
Solution Approach 1:
The picture is divided into multiple areas (e.g., slices, tiles, or rectangular regions) that are encoded and transmitted independently. When packet loss occurs in one area, the error does not propagate to other areas, thus improving reliability while maintaining inter prediction efficiency within each area.
Solution Approach 2:
Different areas of the picture are encoded with different prediction strategies. Areas prone to packet loss can use intra prediction or have their reference pixel usage restricted, while other areas continue to use inter prediction, thus locally optimizing both reliability and efficiency.
2Manufacturing precision
If reference pixels from all areas are used in inter prediction to maintain image quality, then manufacturing precision is improved, but loss of information increases due to packet loss errors
Solution Approach 1:
The encoder identifies and marks areas that are likely to experience packet loss before encoding. This preliminary identification allows the decoder to preemptively avoid using reference pixels from these areas, preventing error propagation and maintaining image quality in unaffected areas.
Solution Approach 2:
The patent converts the harmful effect of packet loss into a beneficial error concealment mechanism. By detecting packet loss and adjusting reference pixel usage accordingly, the system transforms potential quality degradation into a controlled error management strategy that maintains overall image quality.
3Reliability
If intra encoding is applied to all areas to improve reliability against packet loss, then reliability is improved, but productivity deteriorates due to increased encoding complexity and transmission delay
Solution Approach 1:
Instead of applying intra encoding to all areas, the patent applies it only to specific areas where packet loss is likely to occur or where the visual impact of errors would be minimal. This partial application maintains reliability where needed while preserving encoding efficiency in other areas.
Solution Approach 2:
The patent dynamically changes encoding parameters (such as prediction mode, area size, and reference pixel selection) based on content characteristics and estimated packet loss probability. This allows the system to adapt between intra and inter prediction strategies to optimize the reliability-efficiency trade-off.
Data Source
AI summary
An encoder includes circuitry and memory connected to the circuitry. The circuitry, in operation: splits one picture included in pictures into areas, intra encodes the entirety of an area among the areas; and switches areas to be intra encoded for each picture included in the pictures. In at least one of (i) generation of a reference pixel in inter prediction or (ii) loop filtering, only pixels in the area intra encoded or pixels in an area which precedes the area intra encoded in coding order are used among pixels included in the pictures.