Smart Contour Completion for Noisy Scanned Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Scanned engineering drawings often contain unclear tables due to unwanted dots or lines inside cells and incomplete contours, making it difficult to extract cell locations and contents accurately.
Innovation Solution
A method and system for smart contour completion that converts images to grayscale, separates foreground and background regions, extracts horizontal and vertical lines, identifies and removes noisy edges, scans cell boundaries for broken contours, and dilates incomplete contours using a defined kernel size to complete the table image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simple noise removal techniques are used, then processing speed is improved, but incomplete contours remain unresolved and table clarity is not sufficiently improved
Solution Approach 1:
The patent segments the table processing into distinct phases: noise removal, contour detection, join identification, and selective dilation. By dividing the complex task into manageable segments, the system achieves both efficient processing and accurate contour completion without overwhelming computational requirements.
Solution Approach 2:
The patent applies local quality by performing dilation operations selectively only at identified join locations rather than uniformly across the entire table. This localized approach maintains processing efficiency while ensuring precise contour completion exactly where needed, resolving the contradiction between speed and accuracy.
2Manufacturing precision
If uniform dilation is applied across the entire table, then all incomplete contours are completed, but the table structure and existing lines are distorted
Solution Approach 1:
The patent implements local quality by restricting dilation operations to specific join locations identified through contour analysis. Instead of uniform dilation across the entire table, the system applies morphological operations only where contour gaps exist, preserving the overall table structure while completing specific incomplete contours.
Solution Approach 2:
The patent introduces join identification as an intermediary step between contour detection and dilation. This intermediary mechanism identifies precise locations where dilation should be applied, acting as a mediator that prevents unnecessary distortion of the table structure while ensuring complete contour closure at critical junction points.
3Productivity
If aggressive noise removal is used, then processing speed is improved, but legitimate table lines and contours may be removed
Solution Approach 1:
The patent applies parameter changes by adjusting noise removal thresholds based on the specific characteristics of each scanned table image. Rather than using fixed aggressive thresholds that might remove legitimate lines, the system adapts parameters to distinguish between actual table contours and spurious noise, maintaining reliability while preserving processing efficiency.
Data Source
AI summary
The present disclosure recites a system and a method to evaluate the scanned images of tables, identify at least one of the noises, errors and incomplete contours present therein (that make it difficult to extract cell location and it's contains) and process the said images of tables to remove the detected noises/errors from the images of the table and provide final images of table with complete contours. Therefore, said system disclosed herein is configured to take an image of table with incomplete contours (i.e. containing line gaps), as input and provides output in form of a processed image with all contours completed.