Adaptive Early Exit in Image Correlation for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Motion estimation in video compression techniques is computationally intensive, requiring significant resources and increasing costs due to the need for more hardware or complex algorithms to enhance pixel throughput and reduce distortion calculation time.
Innovation Solution
The implementation of an adaptive early exit strategy and partial SAD calculation mode using multiple SAD devices, where the system determines the most efficient calculation method based on image correlation and temporal characteristics, allowing for earlier termination of calculations when sufficient distortion information is obtained.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion estimation uses complete SAD calculation for all search blocks, then measurement precision is improved, but computational intensity and processing time increase significantly
Solution Approach 1:
The patent applies partial action by calculating SAD only for a subset of search blocks rather than all blocks. The system determines a subset of search blocks based on image correlation characteristics and temporal displacement, performing complete SAD calculation only for blocks in this subset while using alternative methods for other blocks, thereby reducing overall computational intensity while maintaining sufficient measurement precision for motion estimation
Solution Approach 2:
The patent applies local quality by differentiating the calculation method based on local image characteristics. Different search blocks are treated differently according to their correlation properties and temporal displacement from reference frames. Blocks with high correlation use simplified methods while blocks with low correlation undergo complete SAD calculation, optimizing the balance between precision and computational efficiency for each local region
2Productivity
If more SAD devices are added to increase pixel throughput, then productivity is improved, but device complexity and hardware cost increase
Solution Approach 1:
The patent applies dynamics by making the calculation method adaptive rather than static. The system dynamically selects between complete SAD calculation and simplified methods based on real-time analysis of image correlation characteristics and temporal displacement. This dynamic adaptation allows the system to achieve high pixel throughput using existing hardware resources without requiring fixed additional SAD devices, thereby improving productivity while controlling device complexity
Solution Approach 2:
The patent applies parameter changes by modifying the calculation approach based on changing parameters such as image correlation characteristics and temporal displacement values. When correlation characteristics indicate similar blocks, the system changes the calculation parameter to use simplified methods; when dissimilarity is detected, it switches to complete SAD calculation. This parameter-based adaptation enables high throughput without proportionally increasing hardware complexity
3Productivity
If early exit strategy is implemented to reduce calculation time, then productivity is improved, but measurement precision may deteriorate due to premature termination
Solution Approach 1:
The patent applies feedback by continuously monitoring correlation characteristics and temporal displacement values during the motion estimation process. This feedback mechanism allows the system to make informed decisions about when to apply early exit strategies versus when to perform complete calculations. The feedback from correlation analysis ensures that premature termination does not significantly compromise measurement precision, as the system can detect when simplified methods suffice and when complete SAD calculation is necessary
Data Source
AI summary
Images are obtained for image compression. The images are compared using sum of absolute difference devices, which have arithmetic parts, and accumulators. The sign bits of the accumulators are determined at a time of minimum distortion between two images. These sign bits are associated with sets of probabilistically-similar parts. When other sets from that set are obtained later, an early exit is established.