Scanned Image Segmentation for OCR Processing Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic document generation systems face inefficiencies due to the need to transmit entire scanned images before processing can begin, leading to prolonged waiting times for OCR processing and document generation, especially when the image includes areas not requiring character recognition.
Innovation Solution
An electronic document generation system that divides scanned images into areas, determines which sections require processing, and transmits only the relevant data early, allowing for concurrent processing and transmission of dummy data to reduce waiting times and improve efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If the external terminal waits to receive all scanned image data before transmitting to the cloud server, then complete image data is available for processing, but the processing start time is delayed and document generation efficiency is reduced
Solution Approach 1:
The scanned image data is divided into multiple divided image data portions. The determination unit identifies which portions contain characters and require OCR processing. Only the necessary divided image data containing characters is transmitted to the cloud server for OCR processing, rather than waiting for or transmitting the entire scanned image. This segmentation enables earlier processing initiation while maintaining reliability by transmitting only the essential character-containing portions.
Solution Approach 2:
The determination unit performs preliminary analysis of the scanned image to identify character-containing areas before transmission. This preliminary action allows the system to prepare and transmit only the necessary divided image data portions to the cloud server in advance, without waiting for complete image reception. The external terminal can then initiate OCR processing earlier while ensuring data reliability through selective transmission of identified character portions.
2Productivity
If the entire scanned image is transmitted to the cloud server, then all image data is available for comprehensive processing, but the communication load and transmission time are increased
Solution Approach 1:
The determination unit extracts and identifies only the divided image data portions that contain characters and require OCR processing. Instead of transmitting the entire scanned image, the system extracts and transmits only these necessary character-containing portions to the cloud server. This extraction principle reduces the quantity of transmitted data while maintaining productivity by enabling faster, more targeted OCR processing on the essential character portions.
Solution Approach 2:
The scanned image is segmented into multiple divided image data portions, and only the segments containing characters are selected for transmission and processing. This segmentation approach reduces the overall data quantity transmitted while improving document generation speed by focusing processing resources on the relevant character-containing segments rather than processing the entire image.
3Measurement precision
If OCR processing is performed on all divided image data, then comprehensive text recognition is achieved, but the processing time and computational resources are wasted on areas without characters
Solution Approach 1:
The determination unit applies local quality assessment by analyzing each divided image data portion to determine whether it contains characters requiring OCR processing. Different portions of the scanned image receive different treatment: character-containing portions are transmitted to the cloud server for OCR processing, while non-character portions are identified and excluded from processing. This local quality approach ensures measurement precision by maintaining accurate character detection in relevant areas while eliminating wasteful processing time in non-character areas.
Data Source
AI summary
An electronic document generation system includes: an image forming apparatus configured to generate a scanned image of an original document; and an external terminal configured to receive image data of the scanned image from the image forming apparatus, and generate an electronic document based on the scanned image, wherein the image forming apparatus includes: a divided data generation unit; a determination unit; and a communication unit, of the plurality of divided image data, the communication unit transmits, to the external terminal, divided image data that is determined to be the processing target data at an earlier point in time, and transmits, to the external terminal, divided image data that is determined not to be the processing target data after the divided image data that is determined to be the processing target data are transmitted, and the external terminal includes: an obtaining unit; and a document generation unit.