Image Block Selection for Time-Limited Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Camera-based retail scanning systems face challenges in real-time image processing due to time constraints and the need to search for product-identifying indicia within composite images, where the pose and perspective distortion of products introduce complexity, exceeding the processing capabilities of common retail scanner hardware.
Innovation Solution
An ordered listing of processing parameters is created, including image excerpt locations and geometrical correction factors, derived from reference scanner images, to prioritize decoding attempts and optimize the scanning system's behavior based on operator habits and product mixes, allowing for improved decoding coverage within the limited processing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the full extent of composite images is searched for decodable indicia, then decoding accuracy is improved, but processing time exceeds available time budget
Solution Approach 1:
The patent divides the composite image into multiple candidate excerpts rather than searching the entire image. This segmentation allows the system to process only relevant portions of the image, reducing processing time while maintaining decoding accuracy by focusing computational resources on likely locations of product indicia.
Solution Approach 2:
The patent applies partial action by examining a limited number of candidate excerpts (e.g., 9 locations) rather than the full image extent. This partial search strategy is sufficient to find decodable indicia in most cases while staying within the time budget, accepting that some edge cases may be missed but overall system performance is improved.
2Difficulty of detecting and measuring
If multiple perspective corrections are applied to compensate for pose distortion, then indicia detectability is improved, but processing complexity exceeds hardware capabilities
Solution Approach 1:
The patent performs preliminary action by pre-computing and storing candidate excerpt locations and their associated perspective correction parameters during an offline training phase. During real-time operation, the system simply retrieves and applies these pre-computed corrections rather than calculating them on-the-fly, significantly reducing processing complexity while maintaining detectability.
Solution Approach 2:
The patent changes parameters by using discrete, pre-determined perspective correction values (e.g., specific tilt and bearing angles) rather than continuous parameter optimization. This discretization reduces the computational burden to a manageable level for retail scanner hardware while still effectively compensating for common product poses.
3Reliability
If a fixed number of image excerpts are examined with multiple perspective corrections, then decoding coverage is improved, but processing speed becomes insufficient for real-time operation
Solution Approach 1:
The patent introduces dynamics by ordering candidate excerpts based on their likelihood of containing decodable indicia, derived from training data reflecting operator habits and product mixes. The system dynamically processes excerpts in this prioritized order, examining more excerpts if time permits, thereby adapting processing depth to available time while maintaining real-time operation.
Solution Approach 2:
The patent uses feedback from training data about operator behavior and product characteristics to inform the ordering and selection of candidate excerpts. This feedback mechanism allows the system to anticipate likely indicia locations based on historical patterns, improving decoding coverage without requiring exhaustive search of all possible excerpts.
Data Source
AI summary
Object recognition by point-of-sale camera systems is aided by first removing perspective distortion. Yet pose of the object—relative to the system—depends on actions of the operator, and is usually unknown. Multiple trial counter-distortions to remove perspective distortion can be attempted, but the number of such trials is limited by the frame rate of the camera system—which limits the available processing interval. One embodiment of the present technology examines historical image data to determine counter-distortions that statistically yield best object recognition results. Similarly, the system can analyze historical data to learn what sub-parts of captured imagery most likely enable object recognition. A set-cover strategy is desirably used. In some arrangements, the system identifies different counter-distortions, and image sub-parts, that work best with different clerk- and customer-operators of the system, and processes captured imagery accordingly. A great variety of other features and arrangements are also detailed.