3D Model Defect Detection Using Silhouette Image Confidence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for generating three-dimensional shape data using synchronous image capturing by multiple imaging devices often result in imperfect silhouette images due to exposure errors, background similarities, and object occlusions, leading to defects like holes and chips in the 3D model, which are difficult to identify and correct, especially when the number of images is large.
Innovation Solution
An information processing apparatus that obtains silhouette images from multiple imaging devices, generates a 3D model, identifies and highlights defect regions, and associates them with the corresponding silhouette images, allowing users to easily detect and modify the defective areas, either by exclusion or automatic regeneration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a plurality of imaging devices perform synchronous image capturing to generate a 3D model, then the three-dimensional shape data can be obtained, but defects such as holes and chips occur in the 3D model due to exposure errors, background similarities, and object occlusions
Solution Approach 1:
The system performs quality assessment of silhouette images by calculating confidence values based on multiple criteria (exposure quality, background complexity, occlusion detection). This feedback mechanism identifies defective silhouette images and enables selective exclusion or correction, thereby improving 3D model quality while maintaining the benefits of multi-device synchronous capturing
Solution Approach 2:
The system performs preliminary quality assessment and defect identification on silhouette images before they are used for 3D model generation. By pre-identifying defective images through confidence value calculation and defect region detection, the system prevents these images from degrading the final 3D model quality
2Area of stationary object
If the number of imaging devices is increased to improve 3D model coverage, then more complete object capture is achieved, but it becomes difficult to identify and correct defective silhouette images
Solution Approach 1:
The system automatically provides feedback by calculating confidence values for each silhouette image and generating quality assessment results. This automated feedback mechanism enables operators to quickly identify defective images even when dealing with large numbers of imaging devices, maintaining ease of operation while improving coverage
Solution Approach 2:
The system introduces an intermediary quality assessment module that acts as a mediator between the plurality of imaging devices and the 3D model generation process. This intermediary automatically evaluates each silhouette image's quality and provides defect information, simplifying the operator's task of identifying problematic images among many devices
3Manufacturing precision
If defective silhouette images are excluded from 3D model generation to improve model quality, then fewer defects occur, but the amount of processing and the number of images to review increases
Solution Approach 1:
The system performs preliminary quality assessment and defect detection on all silhouette images before 3D model generation. By pre-identifying defective images through automated confidence value calculation and defect region detection, the system enables efficient exclusion of only the necessary defective images, minimizing processing time while maintaining high 3D model quality
Data Source
AI summary
To make it possible to easily identify a silhouette image causing a hole or chip having occurred in a 3D model and an imaging device corresponding to the silhouette image. Defect region information identifying a defect region of a 3D model representing a three-dimensional shape of an object, which is generated based on a plurality of silhouette images, is set. Then, the defect region identified by the set defect region information is associated with the plurality of silhouette images and results of the association are displayed.