Watermark Pose Determination Using Reference Data Lookups
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The computational burden of determining the pose of watermarked objects in imagery captured by camera-equipped devices, such as point of sale scanners, is excessive due to processor-intensive methods like Fourier-Mellin domain transformations or iterative least squares approaches, leading to inefficient processing of watermark data.
Innovation Solution
Employing a reference data store and memory lookups, or a convolutional neural network trained with reference data, to determine object pose with reduced computational intensity and time, allowing for efficient watermark processing in applications like point of sale systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If processor-intensive methods like Fourier-Mellin domain transformations or iterative least squares approaches are used to determine pose, then measurement precision of pose parameters is improved, but productivity of watermark processing deteriorates due to excessive computational burden
Solution Approach 1:
The patent pre-computes pose determination data for multiple known pose states and stores them in a reference data store before actual watermark processing. When a watermark needs to be processed, the system simply looks up the pre-computed data corresponding to the detected pose state, avoiding the need to perform intensive Fourier-Mellin transformations or iterative least squares calculations in real-time. This preliminary preparation of reference data resolves the contradiction by trading off initial computational effort for rapid subsequent processing.
2Measurement precision
If processor-intensive pose determination methods are employed, then measurement precision is improved, but loss of time increases due to the computational intensity required for each frame processing
Solution Approach 1:
The patent creates copies of pose determination reference data for multiple predefined pose states and stores them in a reference data store. Instead of performing time-consuming iterative calculations on each captured frame, the system captures the pose state from the watermarked image and performs a rapid lookup of pre-computed pose parameters from the reference data store. This copying and lookup approach maintains measurement precision while dramatically reducing the time loss associated with processing each frame.
3Device complexity
If simple operations are used instead of complex transformations, then device complexity is reduced, but measurement precision of pose determination deteriorates due to the simplified approach
Solution Approach 1:
The patent performs the complex pose determination calculations in advance during an offline phase, storing the results in a reference data store. The online processing phase only requires simple operations such as comparing captured image features with reference data and performing lookups. This preliminary action transfers the computational complexity from the resource-constrained device to a more powerful system, allowing the device to use simple operations while still achieving high measurement precision through the pre-computed reference data.
Data Source
AI summary
A steganographic digital watermark signal is decoded from host imagery without requiring a domain transformation for signal synchronization, thereby speeding and simplifying the decoding operation. In time-limited applications, such as in supermarket point-of-sale scanners that attempt watermark decode operations on dozens of video frames every second, the speed improvement allows a greater percentage of each image frame to be analyzed for watermark data. In battery-powered mobile devices, avoidance of repeated domain transformations extends battery life. A great variety of other features and arrangements, including machine learning aspects, are also detailed.