Learning with Leo: The Cost of Cleaning and Protecting the Ozone Layer
Nolan’s Notes: Cleaning With Smaller Geometries
Standard of Excellence: Building Excellence From the Inside Out
Figure 2: Materials curves (left) generated from training the system on the wedge samples (right).
MACHINE LEARNING/USER INTERFACE
Machine learning algorithms have been developed to enable decisions to be made on material type and thickness based on the materials space plots generated from the image obtained using the MAP. The algorithms require training using a set of training standards. Once the training stage is complete, the algorithm is able to identify the material and thickness of previously unseen samples. Under ideal test conditions, the algorithm has been shown to have a misclassification rate of less than 2% and is able to identify thickness to better than 1% of the true value.
Once the MAP has been fitted to the CMOS detector, the calibration, database training and sample analysis are handled by a simple user interface which is integrated into the X-ray inspection system software. An example is shown in Figure 3, where wedges of three materials are trained and used to identify the material types on the right of the image.
Figure 3: Screen shots showing a user interface which is designed to allow materials training (top) and identification (bottom).
The algorithm works at a rate compatible with image acquisition times and generates a standard grey-scale image as part of the process.
EXAMPLE APPLICATIONS
This article is focused on discussing the X-ray MAP inspection technology for the electronics industry, including PCBs and semiconductor applications. In this section, we show some examples of the MAP X-ray inspection technology applied in the security and food industries. The intention is to enhance the reader’s understanding of the technology and to facilitate the generation of ideas and requirements that can apply for the electronics industry.
Security Inspection
Security threats may be disguised within everyday objects such as laptops and mobile telephones, which are legitimately carried. X-ray security scanners typically use measurements taken at two voltage settings of the X-ray generator in order to generate materials information. This approach requires two scans. Using the MAP, the measurement can be reduced to a single scan at only one voltage setting.
A desk telephone, shown in Figure 4, was measured as an example of a complex object containing electronic circuitry and plastics. Data were collected at 120 kV, 0.5 mA, with a 0.5 s exposure, using a conventional, low-power tungsten X-ray source and a silicon flat-panel detector equipped with the MAP technology. Analysis of the image data leads to the materials discrimination image shown in Figure 4 (right). The color-scheme here is one typically used in security applications: plastics and other organic materials are presented in orange; so called poor metals, such as aluminum, are shown in green; denser metals are shown in blue.
Figure 4: (Left) Absorption contrast image of a telephone. (Right) Materials contrast image showing plastics (orange), poor metals (green) and dense metals (blue).
We see the potential for the same techniques to be applied in PCB inspection to highlight inconsistencies in circuit boards and other electronic components.
To read the full version of this article, which appeared in the June 2017 issue of SMT Magazine, click here.
Page 2 of 2
