Video Frame Rotation for Block Boundary Motion Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression schemes face inefficiencies when encoding and decoding video frames, particularly when objects are located on block boundaries, leading to increased performance costs and artifacts due to suboptimal motion prediction and coding.
Innovation Solution
The method involves rotating video frames by determining the optimal number of degrees for encoding and decoding, changing the raster or scan order to improve performance by altering the object's position relative to block boundaries and enhancing motion prediction, which includes generating prediction residuals based on rotated pixel values and encoding/decoding these residuals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If video frames are encoded using conventional raster scan order, then encoding process is simple, but encoding efficiency deteriorates when objects are located on block boundaries
Solution Approach 1:
The patent applies dynamic scan order adjustment by determining an optimal scan direction (horizontal, vertical, or diagonal) based on the spatial distribution of objects in the video frame. This dynamic adaptation allows the encoding process to optimize for object locations, improving encoding efficiency when objects fall on block boundaries while maintaining simplicity when conventional raster order suffices.
Solution Approach 2:
The patent changes the scan order parameter from fixed raster scan to variable scan directions (horizontal, vertical, diagonal). By selecting different scan directions based on object positions, the system improves motion prediction accuracy and reduces artifacts without requiring fundamentally new encoding mechanisms, thus balancing efficiency improvement with process complexity.
2Productivity
If video frames are decoded using conventional raster scan order, then decoding process is simple, but decoding efficiency deteriorates when objects are located on block boundaries
Solution Approach 1:
The patent implements dynamic scan order selection in decoding by determining the optimal scan direction based on object spatial distribution. This allows the decoder to adapt to different object positions, improving decoding efficiency and artifact reduction while maintaining relative process simplicity through the use of established scan patterns.
Solution Approach 2:
The patent modifies the decode scan order parameter from fixed raster to variable directions (horizontal, vertical, diagonal) based on object locations. This parameter change enables improved motion prediction and artifact reduction without fundamentally altering the decoding architecture, thus achieving efficiency gains with controlled complexity increase.
3Reliability
If conventional encoding methods are used, then processing is straightforward, but motion prediction accuracy deteriorates leading to increased artifacts
Solution Approach 1:
The patent employs dynamic scan direction selection to improve motion prediction accuracy by aligning the scan order with object orientations. This dynamic adaptation reduces artifacts caused by objects on block boundaries while maintaining processing simplicity through the use of standard scan patterns selected based on object positions.
Solution Approach 2:
The patent changes the scan order parameter to optimize motion prediction for different object configurations. By selecting from horizontal, vertical, or diagonal scan directions based on object spatial distribution, the system improves prediction accuracy and reduces artifacts without requiring fundamentally new processing mechanisms.
Data Source
AI summary
Video frames are rotated for encoding and decoding blocks therein. During an encoding process, a number of degrees to rotate a video frame is determined. The video frame is rotated to produce rotated pixel values. A prediction residual is then generated for a block to be encoded based on the rotated pixel values. That block is encoded based on the prediction residual. During a decoding process, an encoded block is decoded to produce inverse transform pixel values, and a number of degrees to inverse rotate an encoded video frame is determined. The encoded video frame is inverse rotated to produce inverse rotated pixel values corresponding to the inverse transform pixel values. The inverse rotated pixel values are then outputted to an output video stream. One or more syntax elements indicating the number of degrees can be encoded to a bitstream during encoding and decoded from the bitstream during decoding.