Hybrid Image Co-processor for Accelerated Machine Vision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing/inspection systems lack sufficient processing power to handle the demands of next-generation manufacturing systems, particularly due to the limitations of general-purpose processors and the inability to offload and accelerate image processing functions at a fine-grained level, leading to increased processing times as data sizes and inspection areas grow.
Innovation Solution
A hybrid computing system is introduced, utilizing a management processor, special purpose engines, a network, and storage devices to distribute and manage image processing tasks, allowing for the reuse of system components and offloading image processing workloads to special-purpose accelerators, enabling accelerated image processing across multiple platforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If all image processing functions are performed within a single general-purpose system, then system simplicity is maintained, but processing power and speed are insufficient
Solution Approach 1:
The system divides image processing functions into discrete, independently executable modules that can be distributed across multiple processors. The image processing system is segmented into acquisition modules, processing modules, and output modules that can operate in parallel on different processors, thereby increasing overall processing speed while maintaining manageable complexity through modular design.
Solution Approach 2:
The patent implements a universal processing framework that can execute multiple different image processing algorithms and functions across heterogeneous processors. The system uses a common interface and standardized data structures that allow the same processing pipeline to operate on different processor types, enabling multi-functionality without proportionally increasing system complexity.
2Productivity
If image processing functions are offloaded to another general-purpose processor, then some processing load is transferred, but no actual acceleration is achieved
Solution Approach 1:
The system changes the processing parameters by utilizing processors with different architectural characteristics optimized for specific image processing tasks. Instead of using identical general-purpose processors, the system employs processors with varying clock speeds, core counts, and instruction set optimizations tailored to different processing stages, thereby achieving actual acceleration while maintaining flexibility through parameter diversity.
3Ease of manufacture
If image processing functions are tied to a specific processor and platform, then implementation is simplified, but fine-grained offloading and acceleration become difficult
Solution Approach 1:
The patent creates a universal processing architecture with standardized interfaces and common data structures that enable the same image processing functions to run on multiple different processor platforms. This universality allows fine-grained offloading of specific processing tasks to appropriate processors while maintaining ease of implementation through a unified programming model and interface specification.
4Area of stationary object
If the inspection area size and gray scale data amount double, then coverage is improved, but processing time increases dramatically
Solution Approach 1:
The system segments the enlarged inspection area into multiple smaller regions that can be processed in parallel by different processors. Each processor handles a specific region or subset of data, allowing the system to maintain constant processing time even as total inspection area doubles, since the workload is divided and executed simultaneously rather than sequentially.
Data Source
AI summary
The present invention relates to machine vision computing environments, and more specifically relates to a system and method for selectively accelerating the execution of image processing applications using a hybrid computing system. To this extent, a hybrid system is generally defined as one that is multi-platform, and potentially distributed via a network or other connection. The invention provides a machine vision system and method for executing image processing applications on a hybrid image processing system referred to herein as an image co-processor that comprises (among other things) a plurality of special purpose engines (SPEs) that work to process multiple images in an accelerated fashion.