Tue, 12 October, 2021
As part of the TACOMA project, Direct Conversion have been working on the development of an X-ray detector and a suite of software libraries to assist with image acquisition and reconstruction. Direct Conversion is a global leader in X-ray detectors for non-destructive evaluation and is the world’s largest manufacturer of cadmium telluride (CdTe) detectors.
TACOMA seeks to develop a system that can offer high-speed detection of all defects expected in automotive CFRP (carbon fibre reinforced polymer) components in the manufacturing process. TACOMA’s primary technique is digital radiography with ultrafast photon counting detectors. All sensors are carried on robotic arms to allow for the inspection of a wide variety of high aspect ratio components.
Direct Conversion has developed the photon counter detector and a range of software modules in order to rapidly collect 2D radiographs of the CFRP components, stitch them together and produce a complete reconstructed image. These include:
- An Actaeon detector: A direct photon counting device
- dll: A C++ based library used to communicate with the detector and allows different acquisition modes to be set up
- An example code: The example code is in C++ and demonstrates how to capture frames from an Actaeon detector using the XcAPI.dll
- IAS: A demo application exhibiting the main features of Direct Conversion detectors including Actaeon, allowing the capture images in different acquisition modes
- An Actaeon simulator: The simulator is a windows console application, following the exact behaviour of an Actaeon detector and greatly facilitates development if the actual hardware is not available
- dll library: Provides a set of tools for the calibration of captured frames
- XcCalibration Application: An application that provides tools to use the XcCalibration.dll
- Image stitching tool: An image stitching application
All X-ray detectors require calibration in order to compensate for pixel-to-pixel or tile-to-tile variations and deliver maximum uniformity over the entire active area. As a result, the calibration procedure incudes a flat field correction feature and can be performed by using the Direct Conversion calibration library (XcCalibration.dll) or a simple calibration procedure.
Figure 1. X-ray image with no object; a) raw image, b) corrected image using a calibration process
Radiographic imaging of components at high geometric magnification will generally mean that the entire component will not fit inside the field-of-view of a single shot from the detector, therefore to obtain a single image of the component an image stitching algorithm will be required. The powerful image stitching process developed for TACOMA includes feature extraction, feature matching, transformation and blending to create a reliable image for the user. In the example below three radiographs, each using 400 frame averaging, have been acquired.
Figure 2. Three radiographs of a set of 400 frame averaging (512x256 pixels) before calibration and stitching is applied
The acquired radiographs have been calibrated and fed to the stitching algorithm and the reconstructed image is shown below. The stitching algorithm works flawlessly and no artifacts to the stitching points are visible.
Figure 3. The reconstructed image (1594x257 pixels) after reassembling the calibrated set of three images