Scan Image Relationship Estimation for Paint Defect Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual inspection of paint defects on vehicle frames is labor-intensive and time-consuming, requiring significant manpower to identify and record defect codes on paint test sheets.
Innovation Solution
A method using machine learning to automatically estimate the relationship between defect sites and codes by recognizing starting points and object codes in scanned images, employing a Convolution Neural Network (CNN) to determine the presence of relationship lines and classify their attributes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual inspection is used to identify and record defect codes on paint test sheets, then defect identification can be performed, but significant manpower and time are required
Solution Approach 1:
The patent replaces the manual mechanical inspection process with an automated image processing system. The system captures images of paint test sheets, automatically recognizes defect codes and their positions, and identifies relationship lines connecting defects to codes using computer vision algorithms, thereby eliminating the need for manual inspection while maintaining high accuracy
Solution Approach 2:
The system enables self-service automation where the inspection process performs its own data collection, recognition, and analysis functions without human intervention. The image processing system automatically extracts defect information, determines spatial relationships, and generates inspection results independently
2Reliability
If manual inspection processes are used, then defect codes can be written and connected with relationship lines, but the process is time-consuming
Solution Approach 1:
The patent implements continuous automated processing where image capture, defect recognition, relationship line detection, and result generation occur in an uninterrupted sequence. The system processes the entire paint test sheet continuously without manual intervention at each step, significantly reducing total inspection time while maintaining reliable defect recording
Solution Approach 2:
The system performs preliminary automated recognition of defect codes and their positions before final defect identification and relationship establishment. By pre-processing the image to locate and recognize all defect codes and their spatial coordinates in advance, the system streamlines the subsequent analysis process
3Productivity
If automated image processing is used to recognize defect codes and relationship lines, then inspection efficiency improves, but complex image processing algorithms are required
Solution Approach 1:
The patent divides the complex image processing task into distinct segments: defect code recognition, defect position detection, relationship line identification, and result generation. Each segment is handled by specialized processing modules that focus on specific aspects of the inspection, making the overall complex system manageable through functional decomposition
Data Source
AI summary
Provided is a method of estimating a relationship between objects through machine learning, the method comprising receiving a scan image, recognizing a starting point of a relationship line and an object code in the scan image, determining whether a relationship line is present between the recognized starting point and the recognized object code, and, based on determining that the relationship line is present, transmitting a combination of the starting point and the object code connected with the relationship line to a database.