Precomputed Pose Lookup for Fast Watermark Decoding in Camera Frames
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 a bottleneck due to the intensive processing required, which has hindered the adoption of digital watermark technology in applications like POS scanners that need to process both barcodes and watermarks within tight time constraints.
Innovation Solution
The use of a reference data store and memory lookups, or a convolutional neural network trained with reference data, to determine object pose quickly and efficiently, reducing computational intensity and time required for watermark processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional pose determination methods (Fourier-Mellin transform or iterative least squares) are used, then measurement precision of pose parameters is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent pre-computes pose determination data for multiple known pose states during an offline training phase, storing these reference mappings in a data structure. During runtime, the system only performs a lookup operation rather than executing complex iterative algorithms, thus resolving the contradiction between precision and computational complexity by shifting the computational burden from runtime to preprocessing.
Solution Approach 2:
The patent creates a simplified copy of the pose determination problem by pre-computing reference data that captures the essential relationship between image features and pose parameters. This reference data structure serves as a compressed representation that enables fast lookup without requiring the full computational machinery of traditional methods during actual pose determination.
2Measurement precision
If traditional pose determination methods are used, then measurement precision is improved, but processing speed decreases due to intensive computation
Solution Approach 1:
The system performs pose determination data computation in advance during an offline training phase, storing pre-computed results in a lookup table. During runtime, pose determination is achieved through fast data retrieval rather than iterative computation, thereby significantly improving processing speed while maintaining accuracy through the pre-computed reference data.
Solution Approach 2:
The patent transitions from static iterative computation to dynamic adaptive lookup by organizing reference data in a structure that allows efficient querying. The system adapts to different pose queries by retrieving pre-computed data rather than re-computing, enabling dynamic response to varying pose determination needs with consistent high speed.
3Measurement precision
If pose determination is performed on all captured frames, then measurement precision is maintained, but productivity decreases due to time constraints
Solution Approach 1:
The patent pre-computes pose determination references offline, enabling fast runtime lookup that can keep up with high frame rates. This allows the system to maintain measurement precision through accurate pose determination while improving productivity by reducing the time required to process each frame, thus enabling processing of all captured frames within time constraints.
Solution Approach 2:
The patent uses a lightweight lookup operation instead of heavy iterative computation for each frame. This disposable, computationally inexpensive approach allows rapid processing of each frame independently, maintaining precision through the reference data while enabling high throughput by minimizing the computational cost per frame.
4Adaptability or versatility
If POS scanners process both barcodes and watermarks, then versatility is improved, but processing time per task decreases due to shared computational budget
Solution Approach 1:
The patent pre-computes pose determination data offline, so that during runtime, watermark processing requires only fast lookup operations. This reduces the time allocated to watermark processing while maintaining dual functionality with barcodes, as the pre-computed reference data enables rapid pose determination without consuming significant processing time from the shared budget.
Solution Approach 2:
The patent creates a simplified reference data structure that captures pose determination logic, replacing complex iterative algorithms with fast lookup operations. This copying approach reduces watermark processing time by using a lightweight data retrieval mechanism, allowing the system to maintain versatility in processing both barcodes and watermarks within the constrained time budget.
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.