AR Marker Recognition Using Adaptive Fixed Threshold Binarization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile terminal devices, such as smartphones, face a trade-off between processing speed and recognition rate when using adaptive thresholding for augmented reality markers, as adaptive thresholding is CPU-intensive and fixed threshold binarization struggles with calculating optimal thresholds within AR markers.
Innovation Solution
An image processing device that combines adaptive thresholding for initial detection of AR markers, followed by fixed threshold binarization using a determined binarization threshold, allowing for efficient recognition and improved processing speed while maintaining a high recognition rate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If adaptive thresholding is used to binarize the captured image, then the recognition rate of AR markers is improved, but the processing speed becomes slow due to CPU-intensive processing
Solution Approach 1:
The patent divides the binarization process into two stages: first using adaptive thresholding to detect AR markers and determine optimal thresholds, then using fixed threshold binarization for the actual marker recognition. This segmentation allows the system to leverage the high precision of adaptive thresholding only where needed (for threshold determination) while using the faster fixed threshold method for the bulk processing, thereby resolving the contradiction between recognition rate and processing speed.
Solution Approach 2:
The patent performs preliminary binarization using adaptive thresholding to detect AR markers and calculate optimal binarization thresholds before the main recognition process. This preliminary action prepares the necessary threshold information in advance, allowing the subsequent fixed threshold binarization to proceed much faster without sacrificing recognition accuracy, thus addressing the speed-precision trade-off.
2Speed
If fixed threshold binarization is used to binarize the captured image, then the processing speed is improved, but the recognition rate becomes low due to difficulty in calculating optimal thresholds within AR markers
Solution Approach 1:
The patent segments the binarization task into two phases: a detection phase using adaptive thresholding to identify AR markers and compute optimal thresholds, and a recognition phase using fixed threshold binarization with the pre-computed thresholds. This segmentation enables the system to achieve both high processing speed (in the recognition phase) and high recognition rate (through optimal threshold determination in the detection phase).
Solution Approach 2:
The patent performs preliminary detection of AR markers and calculation of optimal binarization thresholds using adaptive thresholding before executing the main fixed threshold binarization. This preliminary action ensures that the optimal thresholds are determined based on the actual image content, allowing the subsequent fixed threshold method to achieve high recognition rates while maintaining fast processing speeds.
3Measurement precision
If adaptive thresholding is applied to each frame, then AR markers can be recognized, but the CPU load becomes too high for mobile terminal devices
Solution Approach 1:
The patent segments the image processing into two distinct stages: a lightweight fixed threshold binarization stage for general processing, and a targeted adaptive thresholding stage only when AR markers are detected and need optimal threshold determination. This segmentation dramatically reduces the overall CPU load compared to applying adaptive thresholding to every frame, while still maintaining the capability to recognize AR markers with high accuracy when needed.
Solution Approach 2:
The patent applies adaptive thresholding only partially - specifically only when AR markers are detected and optimal thresholds need to be determined, rather than applying it to all images and frames. This partial application of the more computationally intensive adaptive thresholding method allows the system to maintain marker recognition capability while significantly reducing CPU load and energy consumption.
Data Source
AI summary
The present invention improves the recognition rate of an augmented reality marker and the processing speed thereof, simultaneously. In the present invention, a CPU binarizes actual image captured in an image sensor in accordance with an adaptive thresholding, and detects an augmented reality marker from within the binarized image. Then, the CPU determines a binarization threshold based on the augmented reality marker, and after binarizing the actual image captured in the image sensor in accordance with a fixed threshold binarization method using the binarization threshold, recognizes the augmented reality marker based on the binarized image.