Board Writing Extraction Using Local Threshold Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting board writing from teaching videos often result in unclear, incomplete, or invalid images due to uneven light distribution, varying writing intensity, and unclean writing boards, leading to poor image quality.
Innovation Solution
A method involving target object segmentation, grayscale image conversion with highlighted board writing, and binarization processing, including region block division, weighted sum threshold determination, and noise filtering, to enhance image clarity and completeness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional board writing extraction methods are used, then the extraction process is simple, but the extracted writing is unclear, incomplete, or invalid due to uneven light distribution and varying writing intensity
Solution Approach 1:
The patent divides the writing board image into multiple region blocks (e.g., 8 regions) and performs binarization processing on each region separately using locally adapted thresholds. This segmentation allows each region to be processed according to its specific lighting conditions and writing characteristics, improving extraction clarity while managing complexity through modular processing
Solution Approach 2:
The patent applies different binarization thresholds to different region blocks based on their specific characteristics. Each region's threshold is determined by analyzing the grayscale histogram of that region, allowing local adaptation to varying lighting conditions and writing intensities. This local quality approach improves extraction accuracy without requiring a single complex global processing system
2Reliability
If conventional extraction methods are used, then the processing time is short, but the extracted writing is incomplete or invalid due to noise and poor image quality
Solution Approach 1:
The patent performs preliminary processing steps including converting the image to grayscale, applying target object segmentation to identify the writing board region, and dividing into region blocks before binarization. These preliminary actions prepare the image data in advance, ensuring that the subsequent binarization processing can efficiently and accurately extract complete writing without time-consuming adjustments during extraction
Solution Approach 2:
The patent uses feedback from analyzing the grayscale histogram of each region block to determine appropriate binarization thresholds. The histogram analysis provides feedback about the lighting conditions and writing characteristics of each region, allowing the system to adapt thresholds accordingly. This feedback mechanism improves extraction reliability while maintaining efficient processing through automated threshold determination
Data Source
AI summary
A board-writing extraction method and a related device are provided. The board writing extraction method includes: obtaining a target object segmentation image of a writing-board image, wherein a contrast ratio between a target object and a non-target object in the target object segmentation image reaches a predetermined contrast ratio; converting, according to the target object segmentation image, a grayscale image of the writing-board image into a to-be-processed grayscale image with the board writing being highlighted; and performing binarization processing on the to-be-processed grayscale image to obtain a board-writing image of the writing-board image. The described method and related device can effectively reduce the noise in the board-writing image extracted from the writing-board image.