Parallel Video Disparity Detection Using CPU-GPU Thread Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for testing video quality in customer devices, such as digital television Set Top Boxes, are time-consuming and prone to human error, and can only analyze one device at a time, lacking efficiency in detecting disparities in high resolution video signals.
Innovation Solution
A computer-implemented method utilizing a system with a central processing unit and graphics processing unit to receive and process video signals from multiple devices simultaneously, where CPU and GPU threads work in tandem to detect disparities by converting frames to grayscale, performing image segmentation, and analyzing motion areas for discrepancies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual testing by a person is used to observe video signal, then the testing process is simple to implement, but it is very time consuming and not reliable (prone to human errors)
Solution Approach 1:
The patent replaces manual visual inspection with an automated computer-based analysis system that captures video signals and processes them through algorithms to detect disparities, eliminating human error and time consumption associated with manual testing
Solution Approach 2:
The system performs self-testing by automatically capturing, processing, and analyzing video signals from devices under test without requiring human operators to manually observe and evaluate the video content
2Productivity
If automated analysis of video signal is performed, then testing reliability is improved, but only a single device at one time can be analysed
Solution Approach 1:
The patent segments the video processing task into distinct computational threads - CPU threads for control and coordination, and GPU threads for parallel image processing - allowing multiple devices to be tested simultaneously while maintaining manageable system complexity through modular architecture
Solution Approach 2:
The patent transitions from sequential single-device testing to parallel multi-device testing by utilizing the additional computational dimension provided by GPU processing, enabling simultaneous analysis of multiple video streams without proportionally increasing system complexity
3Productivity
If parallel processing of multiple video signals is implemented, then testing speed is increased, but processing complexity and resource requirements increase
Solution Approach 1:
The patent divides parallel video processing into separate CPU and GPU thread responsibilities, with CPU threads handling device coordination and GPU threads handling intensive image processing operations, thereby increasing testing speed while managing processing complexity through functional segmentation
Solution Approach 2:
The patent merges CPU and GPU processing capabilities into a unified testing system where both processing units work cooperatively on the same video signals, achieving high-speed parallel processing of multiple devices while leveraging the complementary strengths of each processor type
Data Source
AI summary
A computer-implemented method for parallel detection of disparities in video from multiple customer devices under test in a system unit that comprises a central processing unit and is connected to a graphics card with a graphics processing unit, the method comprising: receiving a video signal from each of the devices under test; processing each of the received video signals in a dedicated CPU thread controlled by the CPU and a dedicated GPU thread controlled by the graphics processing unit; wherein within the CPU thread successive threads are uploaded to a memory for the graphics processing unit and the GPU thread is executed to process the frame, and the processed frames are collected from the memory for the graphics processing unit to check within the CPU thread whether disparities have occurred in the processed frame; wherein all the CPU threads are controlled by a main thread controlled by the CPU.