Optical Code Reading With Parallel Inference And Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optical information reading devices face increased processing times due to the heavy calculation load of inference processing by machine learning, particularly in cases where code images are difficult to read, and it is challenging to determine whether serial decoding or inference processing is necessary.
Innovation Solution
The optical information reading device executes inference processing and first decoding processing in parallel using separate cores or threads, with dedicated and general-purpose cores, allowing simultaneous execution of decoding on easy-to-read images without inference processing and inference processing on difficult images, followed by second decoding after inference completion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If inference processing by machine learning is executed to restore appropriate images for reading, then reading accuracy is improved, but processing time increases due to heavy calculation load
Solution Approach 1:
The processor is divided into multiple cores: a first core dedicated to inference processing (restoring appropriate images) and a second core dedicated to decoding processing. This segmentation allows both processes to execute simultaneously without interfering with each other, resolving the contradiction between improving reading accuracy through inference processing and reducing processing time.
2Loss of time
If decoding processing is executed immediately on captured code images without inference processing, then processing time is reduced, but reading accuracy decreases for difficult-to-read images
Solution Approach 1:
The system performs preliminary inference processing on captured code images to generate restored appropriate images before decoding. By preparing the images in advance through inference processing, the system ensures high reading accuracy for difficult-to-read images while the parallel execution maintains efficient processing time.
3Device complexity
If serial execution of decoding processing and inference processing is performed, then processing flow is simple, but total processing time increases
Solution Approach 1:
The processing system is segmented into multiple independent cores that execute inference processing and decoding processing simultaneously. The first core handles image restoration while the second core handles decoding, allowing parallel execution that reduces total processing time without significantly increasing system complexity.
Solution Approach 2:
While the first core is performing inference processing, the second core simultaneously performs decoding processing on other images. This continuous utilization of processing resources eliminates idle time and reduces total processing time, as useful work continues uninterrupted across multiple cores.
Data Source
AI summary
To suppress an increase in processing time due to a load of inference processing while improving reading accuracy by the inference processing of machine learning. An optical information reading device includes a processor including: an inference processing part that inputs a code image to a neural network and executes inference processing of generating an ideal image corresponding to the code image; and a decoding processing part that executes first decoding processing of decoding the code image and second decoding processing of decoding the ideal image generated by the inference processing part. The processor executes the inference processing and the first decoding processing in parallel, and executes the second decoding processing after completion of the inference processing.