Window Image Detection Using Endpoint Directionality and Peak Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional methods cannot accurately and easily convert image files back into CAD drawings, limiting the reusability of design elements like windows, requiring manual redrawing with digital pens.

Innovation Solution

A method and apparatus that extract outlines from drawings, detect end points and their directionality, and identify window images based on peak positions and numbers, enabling accurate detection of window images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional methods are used to convert image files back to CAD drawings, then manual redrawing with digital pens is required, but this process is time-consuming and labor-intensive

Engineering Contradiction:
Improveconversion speedVSAvoidmanual redrawing time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent replaces the mechanical manual redrawing process with an automated image processing system that uses computer vision algorithms to detect window images and extract geometric features from drawing images, thereby eliminating the need for manual digital pen input and significantly reducing conversion time

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables the drawing image to self-identify window elements through automated feature detection and classification algorithms, allowing the image data to serve itself in identifying structural elements without requiring external manual annotation or interpretation

Inventive Principle:
Principle #25Self-service

2Measurement precision

If image files are converted to CAD drawings using conventional methods, then accuracy is compromised, but the proposed method achieves accurate detection

Engineering Contradiction:
Improvedetection accuracyVSAvoiddetection algorithm complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the window detection task into distinct processing stages: outline extraction, endpoint detection, directionality analysis, peak detection, and window identification. This segmentation allows each stage to be optimized independently, improving overall accuracy while managing algorithmic complexity through modular processing

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from analyzing simple 2D image pixels to detecting geometric features in multiple dimensions by identifying endpoints, calculating directionality vectors, and locating peak positions, thereby enhancing detection precision through multi-dimensional feature space analysis

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Adaptability or versatility

If manual redrawing is required for design element reuse, then design element reusability is limited, but the proposed method enables efficient reuse

Engineering Contradiction:
Improvedesign element reusabilityVSAvoidoperation simplicity
Core Design Contradiction:
Adaptability or versatilityVSEase of operation

Solution Approach 1:

The patent creates accurate digital copies of window design elements by detecting and extracting their geometric features from image files, enabling these elements to be copied and reused in CAD drawings without requiring manual redrawing, thereby enhancing design element reusability while maintaining operational simplicity

Inventive Principle:
Principle #26Copying

Data Source

PatentUS10832415B2Window image detection method and device
Publication Date: 2020.11.10 ARCHIDRAW INC
  • US10832415B2 patent drawing
  • US10832415B2 patent drawing
  • US10832415B2 patent drawing

AI summary

Provided are a method and an apparatus for detecting door image, and the method includes extracting an outline from the drawing; detecting an end point of the outline; detecting a directionality of the end point; detecting a peak adjacent the end point using the directionality; and detecting at least one window image using at least one of the number and position of the peak.