Real-Time Font Size Detection in Digital Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting font size in digital images are computation-intensive and inefficient, making it difficult to recognize text in documents with small font sizes in real-time, especially when capturing images with mobile devices, leading to suboptimal user experiences and failed OCR operations.
Innovation Solution
A computer-implemented method and system for real-time font size detection in digital images, which adjusts the image resolution dynamically using a live camera feed, prompts users to zoom in or out, and captures multiple images to ensure legible text, thereby improving OCR accuracy by processing the images in a streaming fashion and stitching them together for analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing font size detection methods are used, then font size can be detected, but the process is computation-intensive and inefficient, making real-time detection impossible
Solution Approach 1:
The patent segments the digital image into multiple smaller regions or chunks, processes each region independently to detect text characteristics, and aggregates results. This division reduces the computational burden on each processing unit while maintaining overall detection accuracy, enabling real-time performance.
Solution Approach 2:
The patent implements a multi-pass processing approach where initial font size detection is performed on a subset of image regions or at reduced resolution, and only regions requiring further analysis are processed in full detail. This partial action strategy reduces overall computation while maintaining detection accuracy for critical areas.
2Area of stationary object
If the entire document is captured in one frame, then the full document is visible, but text with very small font size cannot be recognized
Solution Approach 1:
The patent dynamically adjusts the imaging parameters based on detected font size. When small font sizes are detected in the initial capture, the system automatically triggers a zoom-in operation or requests a new image with adjusted camera parameters, transforming the static capture process into a dynamic adaptive system.
Solution Approach 2:
The system performs initial font size detection on the captured image, uses this detection result as feedback to determine whether the text is legible, and based on this feedback either proceeds with OCR or prompts the user to recapture with better conditions. This feedback loop ensures text recognition accuracy.
3Reliability
If multiple images are captured and processed, then text recognition accuracy improves, but processing time increases
Solution Approach 1:
The patent performs preliminary font size detection on the captured image before committing to full OCR processing. This preliminary action identifies images with insufficient text legibility, allowing the system to request additional images only when necessary, thereby reducing overall processing time while maintaining high recognition success rates.
Solution Approach 2:
The system processes images in a selective manner, applying full OCR processing only to images that pass the preliminary font size and legibility checks. Images that fail the preliminary check are either rejected or used only for guidance, reducing the number of images requiring intensive processing.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to optical character recognition, and more specifically techniques for detecting font size in a digital image. Accordingly to one embodiment, a client device receives a digital image of a document having one or more textual components. The client device finds one or more contours bounding the one or more textual components in the digital image of the document. The client device detects a font size for text contained in the digital image using the one or more contours. The client device extracts the text from the digital image upon detecting that the detected font size is above a defined threshold value.