Image Barcode Detection with Partitioned ROI Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Capturing images at higher resolutions for barcode detection and decoding can improve accuracy but increases computational burden, leading to delays in processing, especially in mobile devices.
Innovation Solution
Partitioning images into sub-images and processing them independently using a detection model to identify regions of interest, followed by decoding, reduces computational load and enhances detection efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If images are captured at higher resolution for barcode detection, then detection accuracy is improved, but computational burden increases
Solution Approach 1:
The patent divides the image processing task into two stages: first, a detection model processes the entire image to identify potential barcode regions; second, only those identified regions are extracted and passed to the decoder for detailed analysis. This segmentation allows the system to maintain high detection accuracy by examining barcode regions in detail while reducing overall computational burden by limiting detailed processing to only necessary areas rather than the entire high-resolution image.
Solution Approach 2:
The patent extracts only the relevant barcode regions from the larger image after detection. Instead of processing the complete high-resolution image through the computationally intensive decoder, the system extracts only the detected barcode areas and their corresponding symbology categories, thereby reducing the amount of data requiring heavy computational processing while preserving detection accuracy.
2Measurement precision
If images are captured at higher resolution for barcode detection, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments the processing workflow into a fast detection phase that operates on the entire image to locate barcode regions, followed by a detailed decoding phase that processes only those specific regions. This time-efficient segmentation prevents the system from performing computationally intensive decoding operations on the entire high-resolution image, thereby reducing processing time while maintaining accuracy through focused analysis of detected regions.
Solution Approach 2:
The patent applies partial action by performing complete detailed decoding only on regions that have been detected as containing barcodes, rather than applying the same level of processing to the entire image. This selective approach ensures that full processing resources are applied only where necessary, reducing overall processing time while maintaining detection accuracy for actual barcode locations.
3Productivity
If the entire image is processed for barcode detection, then all barcodes are detected, but computational resources are overwhelmed
Solution Approach 1:
The patent implements a two-stage segmentation strategy: a detection model first scans the entire image to identify potential barcode locations and their symbology categories, then only those specific regions are extracted and passed to the decoder. This segmentation enables the system to maintain comprehensive detection coverage across the entire image while avoiding the computational overload that would result from applying the full decoding process to every region, as only detected barcode areas receive intensive processing.
Solution Approach 2:
The patent extracts only the necessary barcode regions and their associated symbology category information from the full image for detailed processing. By taking out only the relevant portions that contain detected barcodes rather than processing the entire image through the decoder, the system maintains productivity by detecting all barcodes present while reducing computational resource demand to manageable levels.
Data Source
AI summary
A method includes: capturing an image; partitioning the image into sub-images; for each sub-image: providing the sub-image to a detection model, receiving, from the detection model, one or more sub-image regions of interest (SROIs), each SROI defined by (i) a position of the SROI, and (ii) one of a set of symbology categories, each symbology category encompassing a plurality of barcode symbologies; generating one or more regions of interest (ROI) from the SROIs, each ROI defined by a position of the ROI in the image, and a symbology category; and providing the ROIs to a decoder.