Stereoscopic Aggregation Processor Disparity Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current stereoscopic image processing systems require twice the number of aggregation machines for left-to-right and right-to-left disparity computation, leading to increased power consumption and silicon area, while achieving similar results.
Innovation Solution
A method using a single aggregation processor to calculate and estimate disparity levels for both image capturing sensors by interpolating or extrapolating disparities, applying masks, and calculating confidence penalties, thereby reducing the number of aggregation machines needed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If separate aggregation machines are used for left-to-right and right-to-left disparity computation, then disparity calculation accuracy is maintained, but the number of aggregation machines and power consumption double
Solution Approach 1:
The patent combines the functions of separate left-to-right and right-to-left disparity computation aggregation machines into a single aggregation machine. This is achieved by implementing a unified aggregation process that computes disparity costs in both directions using the same hardware resources, thereby reducing the total number of aggregation machines from two to one while maintaining the accuracy of disparity calculations.
Solution Approach 2:
The aggregation machine is designed to perform multiple functions: it can compute both left-to-right and right-to-left disparity costs, handle different disparity levels, and process multiple image patches. This multi-functionality allows a single aggregation machine to replace what previously required two dedicated machines, reducing system complexity while preserving measurement precision.
2Measurement precision
If separate aggregation machines are used for left-to-right and right-to-left disparity computation, then complete disparity analysis is achieved, but silicon area increases
Solution Approach 1:
The patent merges the silicon area requirements of two separate aggregation machines into a single machine by implementing both left-to-right and right-to-left disparity computation pathways within the same hardware structure. This consolidation reduces the total silicon area occupied by aggregation machines while ensuring that complete disparity analysis is performed through the unified processing architecture.
3Productivity
If multiple aggregation machines are used, then processing capacity is increased, but energy consumption increases
Solution Approach 1:
The patent combines the processing capacity of multiple aggregation machines into a single energy-efficient unit by implementing parallel processing of left-to-right and right-to-left disparity computations within the same hardware. This approach maintains high processing capacity while reducing energy consumption by eliminating redundant computational resources and optimizing the use of shared hardware components.
Data Source
AI summary
A method is provided for use in a stereoscopic image generating system, the system including at least two image capturing sensors and at least one aggregation processor. The at least one aggregation processor is configured to: receive data associated with an image captured by the image capturing sensors; calculate aggregation results for a pre-defined number of disparity levels based on data received from one of the at least two image capturing sensors; estimate aggregation results for data received from another image capturing sensor; and combine the calculated results with the estimated results.