Optical Symbol Detection Using Downscaled CNN Descriptor Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 2D symbol reading systems face challenges in rapidly identifying and accurately differentiating barcodes from varying backgrounds and lighting conditions, with increased processing demands due to higher resolution image sensors, leading to latency and computational power issues.
Innovation Solution
An optical symbol detector using a convolutional neural network (CNN) with a low-density descriptor map, based on S-transform and optimized for SIMD architectures, partitions images into blocks, computes downscaled descriptors, and employs a coarse estimator engine for rapid detection and localization of machine-readable symbols.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If higher resolution image sensors are used to capture more detailed symbol information, then measurement precision is improved, but processing time and computational power requirements increase
Solution Approach 1:
The patent divides the image processing task into multiple stages: initial symbol detection, candidate region identification, and detailed verification. By segmenting the processing workflow, the system can quickly eliminate non-symbol regions without applying full processing power to the entire high-resolution image, thus reducing overall processing time while maintaining detection accuracy.
Solution Approach 2:
The patent applies full computational processing only to candidate symbol regions identified in preliminary detection stages, rather than processing the entire high-resolution image with maximum computational intensity. This partial action approach focuses computational resources where needed, reducing total processing time while maintaining measurement precision in critical areas.
2Productivity
If higher resolution image sensors capture more images per unit time, then productivity is improved, but computational power and energy consumption increase
Solution Approach 1:
The processing pipeline is segmented into lightweight initial analysis stages and more intensive verification stages. By processing multiple images through the lightweight initial stage in parallel, the system can handle higher frame rates with reduced per-image computational power requirements, improving overall productivity without proportionally increasing power consumption.
Solution Approach 2:
The patent extracts and processes only the essential features needed for symbol detection from each image frame, rather than performing complete image analysis on every captured frame. This extraction approach allows the system to maintain high image capture rates while reducing the computational power required for each processing cycle.
3Measurement precision
If traditional full-resolution processing methods are used, then measurement precision is maintained, but processing time and device complexity increase
Solution Approach 1:
The patent implements a multi-stage processing architecture that segments the detection task into coarse detection, candidate identification, and fine verification stages. This segmentation allows the use of simpler processing methods in early stages and reserves complex processing for only those regions that require detailed analysis, reducing overall device complexity while maintaining precision.
Solution Approach 2:
The patent transitions from processing images at full resolution throughout the entire pipeline to a multi-resolution approach where images are processed at different scales and detail levels appropriate to each processing stage. This dimensional change in processing resolution reduces computational complexity in early stages while preserving the ability to achieve high measurement precision in final verification stages.
Data Source
AI summary
An optical symbol detector has an input operative to receive a captured image frame. The detector includes a feature extractor engine coupled to the input to produce a downscaled descriptor map based on the captured image frame, the descriptor map including a convolutional map of blocks of the captured image frame. The detector further includes a coarse estimator engine coupled to an output of the feature extractor engine to produce a detection indication and localization information of an optical pattern indicative of a machine-readable symbol based on the downscaled descriptor map.