Inter Prediction Encoding Using Motion Vector Candidates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compression methods, such as those in MPEG and H.264, face inefficiencies in encoding motion vectors due to increased overhead, particularly when neighboring motion vectors are dissimilar, leading to suboptimal prediction and compression ratios.
Innovation Solution
An inter prediction encoding method that derives spatial and temporal motion vector candidates from neighboring and previously encoded prediction units, selecting one as a predictor to encode the motion vector difference and reference picture index, thereby reducing the amount of coding bits required.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If motion vector information is encoded and inserted as it is, then the motion vector can be accurately decoded, but the compression ratio of image data is decreased because overhead is increased
Solution Approach 1:
The patent creates a copy of the motion vector through prediction using neighboring block motion vectors. Instead of encoding the original motion vector directly, a predicted motion vector is generated by copying and combining motion vector information from adjacent blocks, and only the difference (residual) is encoded and transmitted.
Solution Approach 2:
The patent extracts only the essential information needed for accurate decoding by separating the motion vector into a predicted component (obtained from neighboring blocks) and a residual component (the difference between actual and predicted motion vectors). Only the residual is encoded and transmitted, removing redundant information.
2Loss of energy
If motion vector prediction is performed using neighboring blocks, then the compression ratio is improved, but the encoding complexity is increased
Solution Approach 1:
The patent segments the motion vector encoding process into distinct stages: prediction stage (obtaining motion vectors from neighboring blocks), residual calculation stage (computing the difference between actual and predicted motion vectors), and encoding stage (encoding only the residual). This segmentation makes the complex process more manageable and systematic.
Solution Approach 2:
The patent performs preliminary action by pre-calculating the predicted motion vector from neighboring blocks before the actual motion vector encoding. This preliminary prediction step simplifies the subsequent encoding process by reducing the amount of data that needs to be encoded and transmitted.
3Ease of manufacture
If the median of neighboring block motion vectors is used as predictor, then the prediction can be calculated simply, but the prediction effectiveness is reduced when neighboring motion vectors are dissimilar
Solution Approach 1:
The patent introduces dynamic selection of prediction methods. Instead of always using the median of neighboring block motion vectors, the system dynamically selects between different prediction approaches (such as using motion vectors from specifically positioned neighboring blocks or alternative prediction methods) based on the actual characteristics of the current block and its neighbors, improving prediction effectiveness while maintaining computational efficiency.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Provided is an apparatus for encoding an image that determines a quantization step size of a previous coding block according to scan order as the quantization step size predictor of the current coding block when a quantization step size of a left coding block of a current coding block and a quantization step size of an above coding block of the current coding block are unavailable, divides quantized transform coefficients of a quantized transform block into a plurality of sub-blocks when a size of the transform block is larger than 4x4, and scans the plurality of sub-blocks according to a scan pattern determined by an intra prediction mode of a current block. Therefore, it is possible to reduce the amount of coding bits required to encode motion information and the quantization step size of the current prediction block.