Automatic Foreign Object Detection in Images Using Salient Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image segmentation methods, such as GrabCut, require user interaction and cannot be performed automatically, making them ineffective for detecting foreign objects on background objects in images, especially in complex scenarios like solar panels or semiconductor mask plates.
Innovation Solution
A computer-implemented method that extracts image features, detects a salient region, and segments it using an automatic Grabcut operation without user interaction, generating a region-of-interest (ROI) image and initializing a box to differentiate between foreground and background, employing a Gaussian mixture model for accurate segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional GrabCut segmentation method is used, then segmentation accuracy can be achieved, but user interaction is required which reduces automation and productivity
Solution Approach 1:
The patent applies preliminary action by performing salient region detection before GrabCut segmentation. This pre-processing step automatically identifies and marks the region containing the foreign object, providing an initial boundary that guides the subsequent GrabCut algorithm. This eliminates the need for manual user interaction to define the region of interest, while maintaining segmentation accuracy through the two-stage approach.
2Measurement precision
If manual user interaction is required for segmentation, then segmentation quality can be controlled, but processing time increases and productivity decreases
Solution Approach 1:
The salient region detection step serves as a preliminary action that automatically identifies the foreign object region before GrabCut segmentation. This pre-marking of the region of interest eliminates the need for manual user interaction during segmentation, significantly reducing processing time while maintaining quality through the automated two-stage process.
3Productivity
If automatic segmentation is implemented, then productivity and automation are improved, but segmentation accuracy may deteriorate without user guidance
Solution Approach 1:
The patent divides the segmentation process into two distinct stages: first, salient region detection to identify the rough boundary of the foreign object, and second, GrabCut segmentation to refine the boundary with high precision. This multi-stage segmentation approach maintains accuracy by combining automated region identification with automated refined segmentation, eliminating the need for manual user guidance while preserving segmentation quality.
4Extent of automation
If salient region detection is added before GrabCut, then automation is achieved, but device complexity and processing steps increase
Solution Approach 1:
The patent segments the overall segmentation process into two functional modules: salient region detection and GrabCut segmentation. This modular approach achieves automation by clearly separating the region identification function from the refined segmentation function, making the system more manageable and implementable while maintaining high automation capability.
Data Source
AI summary
A computer-implemented method of detecting a foreign object on a background object in a sample image is provided. The computer-implemented method includes extracting image features of the sample image; detecting a salient region in the sample image; and segmenting the salient region to determine a contour of a foreign object region based on results of segmenting the salient region. Segmenting the salient region is performed by an automatic Grabcut operation without user interaction.