X-Ray Inspection Counting for Overlapping Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing X-ray inspection methods for overlapping objects suffer from decreased accuracy due to reliance on machine learning inference, which can lead to inaccurate object counting when the inference is inappropriate.
Innovation Solution
An X-ray inspection apparatus that combines a learning model with image processing to generate area output information and processed information, allowing for correction of object counts using differences between these outputs, even when the learning model's inference is inappropriate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning inference is used to count overlapping objects, then object inspection accuracy is improved when inference is appropriate, but accuracy decreases when inference is inappropriate
Solution Approach 1:
The patent combines machine learning inference results with image processing results to determine the final object count. The determination unit integrates both approaches, using the learning model's area output information and the image processing's object detection to compensate for each other's weaknesses and improve overall accuracy and reliability.
Solution Approach 2:
The system uses image processing results as a feedback mechanism to verify and correct machine learning inference results. When the learning model provides area information, image processing independently detects objects and compares results, allowing the system to identify and correct inference errors.
2Device complexity
If only a learning model is used for counting, then the process is simplified, but accuracy decreases due to inference omissions
Solution Approach 1:
The patent merges machine learning-based area detection with traditional image processing-based object detection. This combination allows the system to maintain relatively simple processing while improving accuracy, as image processing can identify objects that the learning model may omit.
Solution Approach 2:
The system changes the approach from relying solely on learning model parameters to incorporating image processing parameters as well. By using multiple detection parameters and methods, the system achieves higher accuracy without excessive complexity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Improves the accuracy of object count inspection by correcting for omissions in object identification through image processing, enhancing the reliability of object counting in overlapping scenarios.
Implementation Method 1
an irradiation unit configured to irradiate a plurality of objects with X-rays
Implementation Method 2
a detection unit configured to detect X-rays transmitted through the objects or X-rays reflected by the objects
Data Source
AI summary
An X-ray inspection apparatus includes an irradiation unit configured to irradiate a plurality of objects with X-rays, a detection unit configured to detect X-rays transmitted through the objects or X-rays reflected by the objects, a generation unit configured to generate an image for inspecting the number of objects based on the X-rays detected by the detection unit, and an inspection unit configured to inspect the number of objects based on the image. The generation unit generates area output information obtained using a learning model with the image as an input to identify areas of the objects in the image and processed information obtained by performing image processing on the image. The inspection unit calculates the number of objects based on the area output information and the processed information.