On-device partial recognition for mobile OCR throughput
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current optical character recognition (OCR) systems on mobile devices face challenges in achieving high accuracy due to image quality issues when compressing images for network transmission, leading to reduced throughput and potential errors in data recognition.
Innovation Solution
Implementing on-device partial recognition, which includes performing OCR and barcode recognition on mobile devices to identify and remove non-relevant information, compressing the modified images, and sending them to a server for further processing, thereby reducing image size without compromising quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image compression is applied to reduce file size for network transmission, then throughput is improved, but image quality deteriorates leading to reduced OCR accuracy
Solution Approach 1:
The patent segments the image processing task into two parts: client-side partial recognition (barcode detection, text region identification) and server-side complete recognition (full OCR processing). This segmentation allows the client to perform preliminary processing that reduces the amount of data needing transmission, thereby improving throughput while maintaining accuracy through coordinated processing between client and server
Solution Approach 2:
The patent applies preliminary action by performing partial recognition tasks (barcode detection, text region identification, image segmentation) on the client device before transmission to the server. This preliminary processing reduces the image data size and prepares it for optimized server processing, enabling higher throughput without sacrificing the accuracy needed for complete recognition
2Measurement precision
If high quality OCR is performed to ensure accurate recognition, then measurement precision is improved, but computational power requirements increase reducing throughput
Solution Approach 1:
The patent divides the computational workload between client and server devices. The client handles computationally intensive but simpler tasks (barcode detection, text region identification, image segmentation), while the server performs the more complex complete OCR processing. This segmentation optimizes throughput by distributing computational burden appropriately across devices with different capabilities
Solution Approach 2:
The patent applies partial action by having the client device perform partial recognition tasks that are sufficient for reducing data size and preparing images for server processing. This partial processing approach improves throughput by avoiding the need to transmit and re-process entire high-quality images, while the server completes the recognition tasks that require excessive computational power for client devices
Data Source
AI summary
Disclosed is an approach of on-device partial recognition that includes performing partial recognition on an image of a document captured by a mobile device to detect and/or recognize a specific area (e.g., barcodes, non-relevant text, etc.) and filling the recognized area with a solid color. Because the solid color area has a maximum compression ratio, this approach can lead to image size reduction and increased network throughput for client-server based data recognition where further processing such as advanced data extraction is performed at the server side. The approach can be enforced with neural network algorithms to exclude non-relevant information (e.g., logos, phrases, words, etc.).