OCR Accuracy via Video Frame Stacking and Averaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optical character recognition (OCR) systems face challenges in accurately extracting text from low-resolution or blurry images of documents, often requiring users to provide additional images, which can be inconvenient and reduce efficiency.
Innovation Solution
The use of video frames captured dynamically from cameras, where OCR systems can extract data from successive frames until text is successfully recognized, adjusting camera settings to improve image quality and combining results from multiple frames for accurate text extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If OCR systems process low-resolution or blurry images, then text extraction can be attempted, but recognition accuracy deteriorates
Solution Approach 1:
The system performs preliminary actions by capturing multiple video frames before OCR processing. These frames are stacked and averaged to create a high-quality composite image that serves as the input for OCR, enabling accurate text extraction even when individual frames are blurry or low-resolution
Solution Approach 2:
The system merges multiple video frames through frame stacking and averaging. By combining information from multiple frames, the system creates a composite image with improved signal-to-noise ratio and enhanced text visibility, resolving the contradiction between processing low-quality images and maintaining recognition accuracy
2Measurement precision
If users provide additional images when OCR fails, then text extraction accuracy improves, but operational convenience deteriorates
Solution Approach 1:
The system performs self-service by automatically capturing and processing multiple video frames without requiring user intervention. The frame stacking and averaging process happens autonomously within the application, eliminating the need for users to manually provide additional images when initial OCR attempts fail
Solution Approach 2:
The system maintains continuity of useful action by continuously capturing video frames and processing them through frame stacking until successful text extraction is achieved. This automated continuous process replaces the discontinuous user action of manually providing additional images, improving both accuracy and convenience
3Measurement precision
If multiple video frames are processed for OCR, then text recognition accuracy improves, but processing time increases
Solution Approach 1:
The system performs preliminary frame stacking and averaging before OCR processing, creating a pre-processed composite image that enhances text visibility. This preliminary action reduces the need for multiple iterative OCR attempts, ultimately decreasing total processing time while maintaining high recognition accuracy
Solution Approach 2:
The processing is segmented into distinct stages: video frame capture, frame stacking, averaging, and OCR processing. This segmentation allows optimization of each stage independently and enables parallel processing of multiple frames during the stacking phase, reducing overall processing time while achieving high accuracy through combined frame analysis
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to optical character recognition using captured video. According to one embodiment, using a first image in stream of images depicting a document, the device extracts text data in a portion of the document depicted in the first image and determines a first confidence level regarding an accuracy of the extracted text data. If the first confidence level satisfies a threshold value, the device saves the extracted text data as recognized content of the source document. Otherwise, the device extracts the text data from the portion of the document as depicted in one or more second images in the stream and determines a second confidence level for the text data extracted from each second image until identifying one of the second images where the second confidence level associated with the text data extracted from the identified second image satisfies the threshold value.