Barcode Localization Using Edge Orientation Feature Sets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in rapidly localizing machine-readable indicia, such as barcodes and digital watermarks, within captured images, especially on devices with limited processing capacity, leading to inefficient computation and potential false alarms.
Innovation Solution
The use of machine learning techniques, including feature extraction and classification, to localize 1D and 2D barcodes and digital watermarks by employing edge orientation-sensitive feature sets and probabilistic statistical models, allowing for focused processing on likely areas containing indicia, thereby reducing unnecessary computation and improving detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing technologies are used for localizing machine-readable indicia in captured images, then the localization function is provided, but the processing time is excessive and computation is inefficient on devices with limited processing capacity
Solution Approach 1:
The patent divides the image processing task into distinct segments: first identifying candidate regions using simplified criteria (edges, contours, aspect ratios), then applying full machine-readable indicia recognition algorithms only to these segmented candidate regions. This segmentation approach reduces the overall processing time by avoiding exhaustive search across the entire image.
Solution Approach 2:
The patent performs preliminary actions by pre-identifying candidate regions that are likely to contain machine-readable indicia based on geometric and visual features before applying the computationally intensive recognition algorithms. This preliminary filtering step prepares the data in advance, allowing the main recognition process to focus only on relevant regions.
2Reliability
If existing technologies process the entire image for indicia detection, then comprehensive detection is achieved, but unnecessary computation is performed on areas without indicia
Solution Approach 1:
The image is segmented into candidate regions and non-candidate regions based on visual features such as edge density, contour characteristics, and aspect ratios. The recognition system then processes only the candidate regions, significantly reducing computational energy while maintaining detection accuracy by ensuring all potential indicia locations are covered.
Solution Approach 2:
Different processing strategies are applied to different parts of the image: candidate regions undergo full machine-readable indicia recognition analysis, while non-candidate regions are either skipped or subjected to minimal processing. This local quality approach optimizes energy usage by matching processing intensity to the likelihood of finding indicia in each region.
3Reliability
If traditional methods are used for indicia localization, then the detection function is provided, but false alarms occur and processing efficiency is reduced
Solution Approach 1:
The system performs preliminary analysis of candidate regions using multiple features (edge orientation, aspect ratio, contour characteristics) before committing to a detection. This preliminary action filters out regions that are unlikely to contain valid machine-readable indicia, reducing false alarms while maintaining comprehensive detection coverage.
Solution Approach 2:
The system uses feedback from the preliminary candidate region identification step to guide the subsequent recognition process. Regions that fail to meet the preliminary criteria are excluded from further processing, while regions that pass are subjected to more rigorous verification, creating a feedback loop that reduces false alarms while maintaining detection accuracy.
Data Source
AI summary
The present disclosures relates to finding or localizing machine readable indicia (e.g., a barcode or digital watermark) in imagery. One claim recites an apparatus comprising: memory for buffering blocks of image data, the image data having been captured with a camera and depicting a printed object; one or more processors programmed for: generating an edge orientation sensitive feature set from the image data; using a first trained classifier to determine whether the feature set includes data representing a barcode; and using N additional trained classifiers to determine an orientation angle associated with the barcode, wherein N comprises an integer greater than 3, and wherein the orientation angle is selected based on a probability metric. Of course, other claims and combinations are provided too.