Inter-Frame Prediction Search Templates for Better Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inter frame prediction techniques for video coding fail to fully utilize spatial correlation of adjacent blocks and image frame content correlations, leading to suboptimal compression rates.
Innovation Solution
Implement a method for inter frame prediction that sets a search template based on a block to be predicted, searches for an optimal search block in a reference frame using a designated strategy, and determines prediction information based on the optimal search block, utilizing reconstructed reference regions and decoding strength to enhance the search process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If block-based inter frame prediction is used with motion search strategy, then motion information can be captured, but spatial correlation of adjacent blocks is not fully utilized leading to suboptimal compression rates
Solution Approach 1:
The patent merges the search template with the current block to be predicted, creating a combined search region that leverages both the reference information from the search template and the spatial correlation within the current block. This combination allows the encoder to exploit both temporal (inter-frame) and spatial correlations simultaneously, improving compression efficiency by fully utilizing available spatial information rather than treating them separately.
Solution Approach 2:
The patent performs preliminary construction of the search template using reconstructed reference regions and neighboring blocks before the actual motion search. By pre-processing and preparing the search template with relevant spatial information from adjacent blocks, the system establishes a better starting point for motion estimation, enabling more effective utilization of spatial correlations before the inter-frame prediction process begins.
2Loss of information
If search template is set based on reconstructed reference region and current block, then spatial correlation is utilized, but data transmission requirements increase
Solution Approach 1:
The patent extracts only the essential spatial correlation information from the current block and integrates it with the search template, rather than transmitting the entire current block. By selectively extracting and utilizing only the necessary spatial information (such as gradient information or simplified representations), the system maintains effective spatial correlation utilization while minimizing the additional data that needs to be transmitted.
Solution Approach 2:
The patent employs partial action by using a simplified or reduced representation of the search template and current block combination. Instead of processing and transmitting all possible spatial information, the system applies partial processing that captures the most significant spatial correlations while deliberately limiting the amount of data generated, thus balancing spatial correlation utilization with data transmission efficiency.
3Loss of information
If motion search is performed in reference frame, then inter frame prediction is achieved, but computational complexity increases
Solution Approach 1:
The patent segments the motion search process by dividing the reference frame into multiple search regions based on the search template and motion vector candidates. Instead of performing exhaustive search across the entire reference frame, the system divides the search space into manageable segments, reducing computational complexity while maintaining prediction accuracy by concentrating search efforts in the most relevant regions.
Solution Approach 2:
The patent applies local quality by performing higher precision motion search only in specific local regions where it is most needed, rather than uniformly across the entire reference frame. By identifying and focusing computational resources on critical search areas (such as regions with high motion activity or uncertainty), the system achieves good inter-frame prediction accuracy while significantly reducing overall computational complexity through selective processing.
Data Source
AI summary
The present disclosure may relate to systems and methods for inter frame prediction of a video. A method may include setting, based on a block to be predicted of an image frame, a search template. The method may also include searching for, based on a search strategy, an optimal search block of the search template in a reference frame of the image frame. The method may further include determining, based on the optimal search block, prediction information of the block to be predicted.