Parallel Face Detection via LBP Segmentation on Embedded GPU
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current facial identification techniques face challenges in real-time processing due to high computational demands, especially with high-resolution images and large databases, which limits their application in fields like security where real-time detection and recognition are crucial.
Innovation Solution
A facial identification method utilizing an embedded GPU with multiple processing units to execute face detection and recognition tasks in parallel, optimizing the processing method based on LBP features to improve performance by distributing image processing and face detection across multiple processing units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If facial identification algorithms process high-resolution images with large databases, then identification accuracy is improved, but processing time increases and real-time capability deteriorates
Solution Approach 1:
The patent divides the image processing task into multiple segments by splitting the image into multiple regions or blocks that can be processed independently and in parallel across different CPU cores, thereby reducing overall processing time while maintaining identification accuracy
Solution Approach 2:
The patent transitions from sequential single-core processing to parallel multi-core processing by utilizing the third dimension of time through concurrent execution, allowing multiple image regions to be processed simultaneously rather than sequentially
2Measurement precision
If facial identification algorithms process high-resolution images, then identification accuracy is improved, but computational load increases and processing speed deteriorates
Solution Approach 1:
The patent segments the computationally intensive image processing tasks into smaller sub-tasks that can be distributed across multiple CPU cores, reducing the computational burden on each individual core while maintaining overall processing capability and identification accuracy
Solution Approach 2:
The patent combines the processing power of multiple CPU cores to handle the computational load of high-resolution image processing, merging their capabilities to achieve the required processing throughput without overloading a single processor
3Device complexity
If sequential processing is used for facial identification, then algorithm simplicity is maintained, but processing performance deteriorates when using GPU
Solution Approach 1:
The patent introduces dynamic parallel processing into the previously static sequential algorithm by adapting it to utilize multiple GPU threads and CPU cores, allowing the processing structure to dynamically scale with available hardware resources while maintaining the core identification logic
Data Source
AI summary
A facial identification method includes changing an image size for facial identification, converting the image of changed size to an LBP domain, and detecting a face through scanning across the converted image. At least one or more of steps of converting and scanning is executed by a plurality of processing units.